...making Linux just a little more fun!

July 2006 (#128):

The Front Page

By Heather Stern

Ride the X11 Bus

This one's inspired by Tavis Ormandy of #fvwm on freenode, realising that one of his nearby bus lines to Uni really is the X11 bus. He took a pic to share - I asked him what he thought of my silly idea - the rest, as they say, is history.

You can't really see who's the driver, but hey, most people don't care what the Window Manager is anyway, as long as the keyboard works.

A few passengers aboard the X11 bus: Wilbur, Xteddy. Will there be room to let Konqi aboard?

[BIO] Xteddy - a Gund "Tender Teddy" - was born in 1983, fell in love with Stegu's monitor in the 90's sometime, and since 1998 or so has been faithfully using Unix, though he has been seen on a Windows system now and then and even with a Macintosh once in a while. Lately he has been hanging out a lot in #fvwm on freenode, baking cookies, memorizing people's screenshots so as to be helpful, and indulging in a mocha now and then. The regulars there call him a little hug daemon - a ready source of hugs for all processes. Our Weekend Mechanic is one of his biggest fans.

Our regular readers may recall that Xteddy featured in our second cover art picture over a year ago (back in issue 111). His good pal bear has also stood in for him on occasion.

[BIO] Wilbur has been the GIMP's mascot since whenever it was that he won that logo contest. Best thing he ever did. Now his work gets seen at speaking engagements in just about every major Linux technical conference. He's been working on outreach programs to benighted Windows users, as well.

[BIO] Konqi has been with the K desktop crowd for some time now. He enjoys fleshing out bug reports, and when he really gets into gear, polishing his shiny white taskbar. He and his girlfriend plan to visit Kiagara Falls sometime later this year, and enjoy the purple mists.

For the uninitiated, "Kiagara Falls" is one of the finer wallpapers in the KDE collection. Consider taking a look at kde-look.org.

Talkback: Discuss this article with The Answer Gang

Bio picture Heather got started in computing before she quite got started learning English. By 8 she was a happy programmer, by 15 the system administrator for the home... Dad had finally broken down and gotten one of those personal computers, only to find it needed regular care and feeding like any other pet. Except it wasn't a Pet: it was one of those brands we find most everywhere today...

Heather is a hardware agnostic, but has spent more hours as a tech in Windows related tech support than most people have spent with their computers. (Got the pin, got the Jacket, got about a zillion T-shirts.) When she discovered Linux in 1993, it wasn't long before the home systems ran Linux regardless of what was in use at work.

By 1995 she was training others in using Linux - and in charge of all the "strange systems" at a (then) 90 million dollar company. Moving onwards, it's safe to say, Linux has been an excellent companion and breadwinner... She took over the HTML editing for "The Answer Guy" in issue 28.

Here's an autobiographical filksong she wrote called The Programmer's Daughter.

Copyright © 2006, Heather Stern. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

The Linux Gazette Mailbag


Apache Server Question
OSS external midi plays to AWE
Proofing C
An idea
Speaking of taking over the world...
Followup: [LG#126] mailbag #2
Followup: [LG#127] mailbag #4
Multiple append= Directives in /etc/lilo.conf
Kmail mystery
How to recursively search HTML?
Cannot read Kingston USB pen drive with DOS and Mac partitions
PHP and Apache
Printer setup
Reading Oxygen Phone Manager files
Network Traffic Review/Filtering
Meaning of overruns & frame in ifconfig output
Search Engine Spiders
Profile has errors
Where can I find GLUE (Group of Linux Users Everywhere)
Graham Jenkins
Request for any ADC's driver source code

Apache Server Question

(marnie.mclaughlin at gmail.com) marnie.mclaughlin at gmail.com
Sat May 13 09:17:24 PDT 2006

Answered by: Ben

[ This one slipped through without an answer - anyone out there have a suggestion for Marnie? -- Kat ]

[Ben] - This is from a friend of mine; my memory of how to do this is rather fuzzy, so I'm hoping that someone here will have a bit more recent Apache experience than I do.

Hi Ben,

I have a Linux question for you if you don't mind:)

My boyfriend is a Linux Admin and wants to display a different directory per user based on _SERVER{PHP_AUTH_USER}

For example:


should point to http://server/home/user1/ or

should point to http://server/home/user2/

Both directories should appear to be the same (i.e. /home/) home could be a PHP script or anything else that will work:)

What would you suggest or where can I look for help?

Fank you:)


OSS external midi plays to AWE

sindi keesan (keesan at sdf.lonestar.org)
Fri May 12 18:33:24 PDT 2006

Answered by:

[ No one has had an answered for this one yet. Maybe there's someone reading this issue of LG who does? -- Kat ]

I checked your knowledge base and found my own posting from 2005. The MIDI howto is for ALSA and I am using 2.2.16 kernel and OSS.

Last year you helped me to play AWE files in linux, using my abbreviated version of Slackware 7.1 (I had compiled playmidi and drvmidi but I needed sfxload and a sound bank). I now know how to play FM synthesis and AWE midi in DOS and linux, after setting the sound card in DOS with ctcm and loading soundbank with sfxload. drvmidi or playmidi -a does AWE, playmidi -4 or -f does FM synthesis, and playmidi -e is supposed to play to the external midi device but it DOES NOT.

I also have the file I was missing that was needed to convert rpm to tgz so I can use precompiled rpm packages (rpm2cpio as used by rpm2targz).

If I have inserted awe_wave.o, playmidi -e plays AWE not external, and if I have not inserted it I get an error message about /dev/sequencer not found and device or resource being busy.

cat /dev/sndstat does not do anything (some error message about the device not existing). cat file.midi > /dev/sequencer says the device does not exist or is busy or something.

I have made /dev/sequencer and /dev/midi00 with symlink to /dev/midi and /dev/dsp and /dev/audio and /dev/music and /dev/sndstat.

In DOS I run ctcm to set the addresses and irqs, after which the pnp AWE cards work properly in DOS, then I boot with loadlin into linux and the settings are retained and I can play to external MIDI with several DOS programs, or AWE or FM synthesis.

Soundblaster AWE64 card (ISA, pnp)

Using a 2.2.16 kernel which I compiled with support for sound:

insmod soundcore
  insmod soundlow
  insmod sound
  insmod v_midi (Do I need this one? Is it causing problems?)
  insmod uart401
  (Sometimes also insmod mpu401, makes no difference)
  insmod sb io=0x220 mpu_io=0x330 irq=5 dma=1 dma16=5
  insmod awe_wave (for AWE sound only)
  sfxload /usr/lib/sfbank/synthgm.sbk (the sound bank)
  insmod opl3 io=0x388 (for FM synthesis only)

Or with a 2.4.31 kernel

  insmod soundcore
  insmod sb_lib
  insmod sb.....

Do I need support for joystick or gameport as well? My kernel has modular support for joystick but I don't have a joystick.

The only info I could find online was requests from other people who had their external midi devices working in Win95 not in linux. With playmidi -e or Rosegarden they either did not play, or played as FM synthesis on a card without AWE. Mine plays as AWE, perhaps because I set that as the playmidi default.

External midi works in DOS with several programs such as playb.exe (playpak) and beatmaster. We have an OEM midi cable and a Yamaha Clavinova with the cable plugged in properly (OUT of cable to IN of piano and vice versa - no sound came through when we plugged it backwards). The joystick port is not disabled on the card in ctcm, and I tried a non-AWE card too with jumper set to joystick enabled.

I also tried the basic (bare.i) kernels from Slackware 7.1 and 10.2 and I think 8.1 (one of them would not even play AWE sound). They all seem to have full sound support, mostly as modules.

What little step might I have left out, or is there a bug in playmidi?

You helped me back in around 2002 to set up a 2MB ramdisk linux with mdacon (to use both TTL and VGA monitors) and tried to teach me how to use screen and suid. I still do not have modprobe, just insmod.

Proofing C

The Answer Gang (tag at linuxgazette.net)
Wed Apr 5 14:03:27 PDT 2006

Answered by: Ben, Jason, Neil

[ This was originally part of the Talkback: 122/sreejith.html discussion. -- Kat ]

[Jason] - [ Aside to the Answer Gang: I am by no means a C expert, but I know a little bit about the language. ISTR some reader volunteering to proofread C code that comes up in future articles, but if not, I could step in and try to catch the obvious technical stuff.]

[[Ben]] - Well, Jan-Benedict Glaw volunteered - but I haven't actually seen any result from it, despite prodding him a couple of times. I guess he's been busy, or something. Your offer is gratefully accepted, since my C is too rusty to be useful except in trivial cases.

[[[Neil]]] - Feel free to CC me for a second opinion on any C/C++ stuff.

[[[[Ben]]]] - That's great, Neil; accepted with thanks. Even with Jason's kind offer, it would still be very nice to have more than one person available to do this.

Side note: anyone who contributes to the proofing process - technical, HTML vetting, whatever - please add your name to that issue's STATUS file as described there. We do not yet have a mechanism for using this, but at some point in the near future, I'm hoping to have a bit of processing added that will display those names on the index page of each issue. Either via my ham-handed efforts, or someone else's Python skills (which would be greatly preferred; when I take an axe to Python, it ain't pretty), but it will happen.

An idea

Benjamin A. Okopnik (ben at linuxgazette.net)
Wed Apr 19 18:23:07 PDT 2006

Answered by: Ben, Pablo, Thomas
[ cc'd to The Answer Gang for comments and further suggestions ]
On Wed, Apr 19, 2006 at 05:49:17PM -0700, Pablo Jadzinsky wrote: 
  > Dear editor, 
  > I am somewhat newbie to linux and I find your magazine the best  
  > resource I found so far. Even though I have great pleasure reading it  
  > and I find it extremely useful I have a suggestion. 
  > I think it would be great to have a 'homework' forum where every  
  > month a script or task can be suggested and then we (the readers)  
  > work on in during say 2 weeks. Some of us can send solutions to you  
  > and then someone on your team can post some of the different  
  > solutions with perhaps some comments on the different approaches. My  
  > experience says that the only way of really learning linux is through  
  > practice but at the beginning is difficult to use the system altogether. 
  > Any way, I hope you find my comment appropriate and useful. 
  > Thanks a lot 
  > Pablo Jadzinsky 

Pablo, that's an excellent idea. I like it a lot; in fact, this is one of the things I tried to do when I wrote the "Learning Perl" series here a while ago. In fact, your idea gives me an idea:

How would you like to write this column?

Before you start objecting due to the fact that you're a "newbie" [1], consider these two facts:

1) You don't have to come up with an answer to them yourself (although I certainly hope that you'll try); posting them to The Answer Gang should get a nice variety of answers which you could collate and use as part of next month's article.

2) I suspect that most of us (it is certainly true for me) have by now forgotten the specific kinds of problems that new users face, and would have a tough time coming up with a list of them that would be relevant. You, on the other hand, have the advantage of your position as a new user, and can simply use any problems that you run into (after considering if they are of general interest, of course.)

Besides... I don't like to mention it, because it's sort of like twisting people's arms... but think of the *fame*. I mean, just imagine - "Pablo Jadzinsky, Linux Gazette columnist." In *bold*, even. Women will swoon, men will grit their teeth in envy, little green creatures from Beta Aurigae will finally make contact with us because we've shown ourselves to be so highly evolved... it's a winning scenario all around. Really. :)

[1] To quote Tom Christiansen, whose opinion, I fully share,

  I find "Newbie" to be a grating, overly-cutsie buzzword 
  (read: obsequious neologism) used by people who are either
  trying to be overly-cutsie or else who don't know that 
  "neophyte", "beginner" or "I'm new to" work just fine. 
  It sounds like something you'd be likely to find 
  in that offensively entitled line of books called 
  "XXX for Dummies".
[[Thomas]] - Been done before via various means --- some people (including myself) tried setting various exercises in articles. But then this is only targetting a specific audience: the people reading that specific article. (Ben's Perl series at least had some reader feedback in this way).

[[[Ben]]] - Speaking of titles - "Thomas Adam, Linux Gazette cynic-in-residence." Think of the fame, the... oh, right. :)

My articles were intended as lessons in basic Perl for those who wanted an easy start. The exercises were purely incidental. I believe that a column of the sort that Pablo suggested would garner an audience over time - it may even get a good jump-start right away.

[[Thomas]] - Sometimes allusions were made in the earlier editions of TAG for readers' to try out ideas if they were so inclined, and do post their results in. Some did, but not many.

It's worth a try, for sure. I am greatly cynical though, so do not be surprised if the response you get from it is low at best. You never know, you might get a 'good' month in it.

[[[Ben]]] - Why not leave off the discouraging grumbling, and see what happens? Thomas, there are no positive outcomes to be served by carping in this case, and lots of negative ones. Why do it?

[[Thomas]] - I assume (since it's your suggestion) you'll be detailing to us in more detail the sorts of things you were meaning, with an example? Yes? Excellent.

[[[Ben]]] - In fact, if Pablo does decide to undertake this, I would rather he didn't detail them. Granting him the same courtesy that we do to any author means leaving the details to him up until the moment that he submits the finished article. There's no reason whatsoever to demand them until that point.

[[[[Thomas]]]] - I wasn't saying that. From what Pablo is saying, I get the impression that he would like us at TAG to critique the answers?

[[[[[Ben]]]]] - Actually, that was my suggestion. I saw these two birds, and I only had the one stone...

[[[[Thomas]]]] - If that's so, then that sounds like a good idea to me (is this positive enough?) I'm just curious, that's all. Heck, it's an interesting discussion.

[[[[[Ben]]]]] - You bet. :) Encouraging participation is a plus in my book.

[[[[[[Pablo]]]]]] - I can't believe I got you two to write this much. If I tell my wife that in top of doing my Ph.D, changing diapers and starting 2 companies I am going to be involve in writing a column she will divorce me right away. I am seriously tempted but let me see what I
can seriously manage and I'll get back to you.

By the way Ben, your intro to scripting in bash is excellent.

Speaking of taking over the world...

Benjamin A. Okopnik (ben at linuxgazette.net)
Tue May 2 06:41:53 PDT 2006

Answered by: Ben, Martin, Suramya, Thomas
"Gee, Brain - what are we going to do tonight?"
  "The same thing we do every night, Pinky - try to take over the world."

So, my nefarious plan is this: let's all write a column (yep, I've definitely caught the bug from the idea proposed by that querent, WhatsIsName, recently.) What I'm talking about is creating a monthly thread in which we walk through a common Linux task - anybody is welcome to toss in the seminal idea and the first part of it (I'll take dibs on 'Installing a printer' this month :), and the rest of us can pitch in comments: different ways to do it, solutions to problems that could come up during the process, hardware/software quirks to watch out for, etc. At the end of each month, I'll format the whole thing into an article - I still need to figure out the layout, perhaps something highly structured with lots of footnotes - and the credit will go to all the members of TAG who participated.

What do you all think?

[Thomas] - I think it's a nice idea -- and something that's worth having a go at. ;)

[[Ben]] - A pleasure to see you positive and first out of the gate, Thomas. :) Very cool indeed.

[Martin] - That would be a interesting idea... Certainly would like to know more about modern day printing...

All I have to do with the distro I use is select a printer and it sets it up. I presume I'm using Cups although I'm not too sure...

[[Ben]] - Well, that is the way it often works - but that's not something from which you can learn a whole lot. However, I'm more interested in the times when it doesn't work that way; I've run into a lot of those over time, and have learned quite a bit out of that. I'm definitely hoping that other people can contribute something from their less-than-perfect experiences as well.

[[[Martin]]] - Luckily every distro I've tried has drivers for my printers but I still would like to find out how it all works without...

[Suramya] - Sounds like a great idea.

Followup: [LG#126] mailbag #2

Jimmy O'Regan (jimregan at o2.ie)
Tue May 2 12:07:27 PDT 2006

Answered by:

Following up to a Mailbag item from issue #126:

GnomeMeeting is now called Ekiga: http://www.ekiga.org .

Followup: [LG#127] mailbag #4

Ganesh Viswanathan (gv1 at cise.ufl.edu)
Thu May 25 19:02:29 PDT 2006

Answered by: Ben, Neil, Thomas

[ This is a followup to the question I have a question about rm command. from LG#127. -- Kat ]

My reply for the:

"I have a question about rm command. Would you please tell me how to remove all the files excepts certain files like anything ended with .c?" question:


The simple command (bash)

  rm *[!.c]

would work for deleting all files except the right?

For all directories also, one can use:

 rm -rf *[!.c]



[Thomas] - Assuming you had:

  shopt -s extglob


  > For all directories also, one can use: rm -rf *[!.c]

I'd use find, since the glob will just create you unnecessary headaches if the number of files are vast.

[[Ben]] - That would be my solution as well. In fact, since I'm a fan of "belt and suspenders"-type solutions for shell questions of this sort ('rm' is quite ungracious; it does exactly what you tell it to do - I'd have sworn I've heard it chuckle at me...), I'd use 'find' and 'xargs' both.

[Neil] - [Quoting Ganesh]:

   > The simple command (bash)
   > rm *[!.c]
   > would work for deleting all files except the right?

That doesn't look right to me. I would expect that to delete all files, except those ending in '.' or 'c', so it would not delete example.pc or example.

   > For all directories also, one can use:
   > rm -rf *[!.c]

That will probably delete far more than you want. The glob expression is evaluated by the shell in the current directory. The rm command does not see the expression "*[!.c]", it sees whatever list the shell has created from that expression, so if you had a directory containing readme.txt, and subdirectories src and module1, rm will get the command line "rm -rf readme.txt module1" and will delete everything in module1, including files ending in .c. The subdirectory src won't be matched, because, as I said above, it doesn't match anything ending in 'c'.

Here's a little demo:

  neil ~ 07:51:14 529 > mkdir /tmp/test
  neil ~ 07:51:26 530 > cd !$
  cd /tmp/test
  neil test 07:51:31 531 > mkdir src
  neil test 07:51:38 532 > mkdir module1
  neil test 07:51:45 533 > echo -n > readme.txt
  neil test 07:51:52 534 > echo -n > example.pc
  neil test 07:51:58 535 > cp readme.txt example.pc src/
  neil test 07:52:06 536 > cp readme.txt example.pc module1/
  neil test 07:52:09 537 > ls -lR
  total 8
  -rw-r--r-- 1 neil users 0 May 26 07:51 example.pc
  drwxr-xr-x 2 neil users 4096 May 26 07:52 module1
  -rw-r--r-- 1 neil users 0 May 26 07:51 readme.txt
  drwxr-xr-x 2 neil users 4096 May 26 07:52 src

  total 0
  -rw-r--r-- 1 neil users 0 May 26 07:52 example.pc
  -rw-r--r-- 1 neil users 0 May 26 07:52 readme.txt
  total 0
  -rw-r--r-- 1 neil users 0 May 26 07:52 example.pc
  -rw-r--r-- 1 neil users 0 May 26 07:52 readme.txt
  neil test 07:52:16 538 > rm -rf *[!.c]
  neil test 07:52:27 539 > ls -lR
  total 4
  -rw-r--r-- 1 neil users 0 May 26 07:51 example.pc
  drwxr-xr-x 2 neil users 4096 May 26 07:52 src
  total 0
  -rw-r--r-- 1 neil users 0 May 26 07:52 example.pc
  -rw-r--r-- 1 neil users 0 May 26 07:52 readme.txt
  neil test 07:52:30 540 > 

Multiple append= Directives in /etc/lilo.conf

moped (moped at sonbeam.org)
Thu Apr 6 10:13:41 PDT 2006

Answered by: Ben

I found a web page with your subject comment, but I couldn't find where you actually explained how to do it. Can you put a per-image append kernel option in lilo.conf? Or does something like that have to be done all on one line, as you suggest?

Could you please give me more info on how to do this?


[Ben] - Sure; here's a copy of the relevant part from my own "lilo.conf":

# Set the default image to boot

        append="quiet acpi=on idebus=66 hdc=ide-scsi"

        append="quiet acpi=on idebus=66"

        append="resume=/dev/hda2 quiet acpi=on idebus=66 hdc=ide-scsi"

Each image entry gets its own 'append' option. This is also documented in the LILO documentation; under Debian, the '/usr/share/doc/lilo/Manual.txt.gz' file contains several examples as well as an explanation of the option.

Kmail mystery

Neil Youngman (ny at youngman.org.uk)
Fri Apr 28 01:00:48 PDT 2006

Answered by: Neil

[ Neil found his problem immediately, but I thought that the TAG readership would like to see his nicely clueful help request, anyway. -- Kat ]

I noticed this morning that kmail was sending messages via postfix on my machine, not via my ISP's mail server and as a consequence some emails weren't being delivered because it's a dynamically allocated IP. this was unexpected because

a. I thought kmail was set up to deliver via my ISP

b. postfix was set up for local delivery

First off I reconfigured postfix (dpkg-reconfigure postfix) to make sure that it was configured for local deliveries only. This made no difference that I could see.

I then checked my kmail configuration. Under the accounts/sending tab, I saw that none of my sending accounts were marked as the default. That was unexpected, but seemed like a reasonable explanation for the problem. I selected the sending account for myisp and retested. It still sent via my local postfix program.

I took a closer look and that account had no hostname set. I'm starting to wonder how my kmail settings got into this state. I fix it and retest. It still goes via my local postfix! I stop and restart kmail, check the settings and retest. No change.

I am now completely at a loss. I search through all the visible kmail settings and I can't see any other settings that should affect how my email is sent.

My default sending account has the hostname for sending email set to mail.myisp.com. I can telnet to the SMTP port at mail.myisp.com, but kmail insists on sending mail to localhost. Why? at am I missing?

[Neil] - I am mistaken. The last 2 emails did go via my ISP, so somewhere along the line kmail picked up the change and I misread the logs.

How to recursively search HTML?

(m at de-minimis.co.uk) m at de-minimis.co.uk
Wed May 17 20:13:40 PDT 2006

Answered by: Ben, Francis, Thomas

Dear LinuxGazette,

Can you recommend a tool for recursively searching for a a given word starting at a given HTML page?

Google's great when on the web, but when I have HTML on my local machine Google can't get to it.

It should be reasonably easy to script something together using wget, awk (to get rid of anything that's not meant to be seen), and grep, however if there are already nice solutions out there one might as well use those.

Best Wishes, Max

[Thomas] - Do you just mean the content? Open it up in a browser. If it's the actual HTML, use an editor.

[Francis] - If you want to have content served by your web server searchable on an ongoing basis, then the best answer is probably to run a local search engine, such as htdig or swish-e (or, presumably, many others).

If you want content on my web server to be searchable, I'd prefer if you didn't fetch everything without asking me first.

"web server" there is for three things: get at the content initially for indexing; provide an interface to the search utility for searching; and allow the search utility present a link to the original content for retrieving. None of the three requires a web server, but for search engines which expect to be used on a web site you may need to make non-default configurations if you want your primary interface to be file: urls instead of http: ones.

  > It should be reasonably easy to script something together using wget, awk 
  > (to get rid of anything that's not meant to be seen), and grep, however 
  > if there are already nice solutions out there one might as well use those.

If it is to be a one-off or rare search, then "find | xargs grep" on the filesystem should work. The equivalent through the http interface would also work. ("find" becomes "print a list of urls"; that then would go through a url-retriever like wget or curl to spit the content to stdout for grep; if grep matched, you'd want to print the matching url for later consideration.)

In either case, you'll effectively be searching all of the content every time.

You can get your search engine to search all of the content once, and then tell you where to look for specific terms. The larger the volume of content, the sooner it will have been worth indexing it initially.

[Ben] - I suppose it depends on what you mean by "recursively" - for many webpages, enough recursion depth means searching the entire web.

I don't know of anything like that already made up, but as you say, it should be easy enough to script. My first cut would be something like 'wget' piped into 'lynx -dump -nolist' plus 'grep' to filter the output - but YMMV.

Cannot read Kingston USB pen drive with DOS and Mac partitions

Sindi Keesan (keesan at grex.cyberspace.org)
Sun May 14 10:15:13 PDT 2006

Answered by: Thomas

Someone asked for our help repartioning/reformatting (to FAT16) a Kingston DataTraveler Elite 256MB USB 2.0 pen drive which currently contains one DOS partition of about 200MB that can be accessed via a Mac (which was used to delete the files in it), and a 50MB or so Mac partition that cannot be reformatted by the Mac. (Something is greyed out).

The drive was apparently given out by a TCF bank at a conference and one of the partitions is labelled TCF_2005.

usb-storage.o identifies it as Model: TCF_2005 and Type: CD-ROM

I was told that DOS USB drivers also found it as CD-ROM, and Windows XP or 2000 found DOS and Mac partitions. Our DOS drivers did not find it at all on our older computers. We have only USB 1.0 ports.

I attempted to use the Slackware 10.2 bare.i kernel with hfs.o (Mac file system) module, which has no dependencies listed in modules.dep.

I use Slackware-4.0- and uClibc-based Basiclinux 3.40 and Slackware-7.1-based Basiclinux 2.1 with glibc upgraded to 2.2.5.

When I insmod hfs.o:

Unresolved symbol: generic_file_llseek

(copied via pencil and paper - how do I save such messages to a file?).

hfsplus.o gave twice as many lines of messages.

I found only four references to the first three lines in google/linux and they were for later kernels.

The drive comes with Windows or OSX security software which I suspect is causing the problem.

I cannot use fdisk without first finding the drive as a device.

Is it possible to repartition this drive with any linux?

Sindi Keesan

[Thomas] - Why insmod? That's not really the best way to go. You almost certainly want to use modprobe so that any dependant modules needed by the one you're trying to load, do so.

  > (copied via pencil and paper - how do I save such messages to a file?).

They'll appear in /var/log/messages.

  > hfsplus.o gave twice as many lines of messages.

I presume with the same content as the above?

  > I found only four references to the first three lines in google/linux 
  > and they were for later kernels. 
  > The drive comes with Windows or OSX security software which I suspect 
  > is causing the problem. 
  > I cannot use fdisk without first finding the drive as a device.

Such things will appear as mass storage devices. Usually via hotplug, they'll be assigned one of the /dev/sda* mappings. In fact, hotplug is something that is both useful and a PITA. For kicks, try resetting it, removing your pen drive first, and then reinserting it:

  sudo /etc/init.d/hotplug restart 

(if you don't know what sudo is, or don't have it setup, you'll have to su to root to do the above.)

You should look at the output from 'dmesg' to see what else it says about your device. Something you haven't even bothered to tell us (but that's OK -- everything is about guessing around here) is the kernel version you're using. Guess what mass storage devices use? Go on -- you might even get a gold star out of it. No? Ok, it's SCSI. In the 2.4.X kernels, you'll need some sort of emulation layer to do this -- scsi_mod probably.

If you're in 2.6.X, you don't have to worry about that.

> Is it possible to repartition this drive with any linux?

It is -- but this is 200MB we're talking about here...

PHP and Apache

clarjon1 (clarjon1 at gmail.com)
Fri May 12 12:29:22 PDT 2006

Answered by: Ben, Thomas

Hey gang, hope you can help

I am running Slackware 10.2 with the default apache and php stuff that came with the installation. I am having a few troubles with PHP. I have done what I have read I am supposed to do, the modules are loaded, but when I goto view a PHP doc via the server, I get either a blank page or a Download To box (depending on the browser, and what page it is) Any suggestions? I would really appreciate.


(PS, I can't wait to finish writing my first article to submit to LG! :) )

[Thomas] - [Quoting clarjon1]

 > I have done what I have read I am supposed to do, the modules are

Really? And just what have you read?

You can't have done everything correctly. This symptom is usually the result of Apache not knowing how to deal with that file type. I am going to make an assumption here in that you really have got the PHP modules loaded. Of course, since you don't tell us which version of Apache you're using, it makes things slightly harder. I don't use Apache2 (way overhyped), so for Apache 1.3.X, you have to ensure in your /etc/apache/httpd.conf file, you have something like:

 AddType application/x-httpd-php .php 
 AddType application/x-httpd-php-source .phps

[[Clarjon1]] - Sorry for not specifying what all the versions are, I was at school at the time, and was pretty rushed for time.

[[[Thomas]]] - It happens.

[[Clarjon1]] - I have Apache 1.3.33, and I have those modules loaded in both mod_php and httpd.conf, so I actually get errors about the modules already loaded,

[[[Thomas]]] - This is normal, and quite harmless. Which version of PHP are you using?

[[Clarjon1]] - The AddType lines, I have those added into the httpd.conf file, and the application/x-httpd-php stuff in the mimetypes file as well (without the AddType there, of course)

[[[Thomas]]] - Right.

[[Clarjon1]] - The module is there, I think that it could be that I messed something up somewhere in installation, and it's biting me now. I'm gonna try downloading fresh copies of apache and php later today, and compile from the source when I get home.

[[[Thomas]]] - Possibly overkill. It's probably not the applications per se that are causing your issue, but one of misconfiguration. I would not have said that recompiling anything is going to help, but YMMV.

[[Clarjon1]] - One last question, what sort of errors should I look at in the error/access logs for hints as to what is going wrong? Thanks for taking the time to help, guys. I really appreciate this.

[[[Thomas]]] - There probably won't be any errors, since Apache is doing exactly what you asked it to do (i.e. with a misconfiguration, or no configuration, it will attempt to either display the php file, or give you the option of downloading it). I actually have a multitail window open (via sshfs) that tails both /var/log/apache/{access,errors}.log:

  multitail -I /mnt/var/log/apache/errors.log -I \

It's quite useful for things like this. :P

[[[[Clarjon1]]]] - Thanks for all your help, but I was able to fix the problem. Turns out I had all the configuration files set up correctly, I was missing some files. I don't know why nothing complained, but oh well. I went through the packages from my other distros I have kicking around, and I found this package: libphp_common432

This package provides the common files to run with different implementations of PHP. You need this package if you install the php standalone package or a webserver with php support (ie: mod_php).

I installed it, and viola! I now have PHP working. I got a little forum set up (after finding mySQL was broken and then had to install postgres, then figuring out how to use postgres and the database and users which can be used via web). I'm glad that's over with.

So all is good for now, with this problem Thanks again for your help!

[Ben] - As Thomas said, it sounds like you don't have the PHP filetypes defined in your 'httpd.conf' - or perhaps the relevant module is missing or not enabled. On Debian, this happens automatically as part of the PHP package installation procedure; presumably, Slackware doesn't do that, so you'll have to do it manually. In my /etc/apache2/apache2.conf, the relevant entries look like this:

  DirectoryIndex index.html index.cgi index.pl index.xhtml index.php
  Include /etc/apache2/mods-enabled/*.load

I'd also ensure that there was a 'php5.conf' and a 'php5.load' in my /etc/apache2/mods-enabled/ directory. For plain Apache (i.e., not
Apache2), the above would be more of a direct reference from 'httpd.conf' and would look something like

  AddType application/x-httpd-php .php
  AddType application/x-httpd-php-source .phps
  LoadModule php5_module /usr/lib/apache/1.3/libphp5.so  

(This would, of course, require that the module named above existed at that location.)

Something I've also found helpful in my PHP travails of the past (this wouldn't help in your case, since the problem you're having comes before this level of troubleshooting) is having a file called 'info.php' living somewhere in my web hierarchy and consisting of the following:


Once PHP is up and running, looking at this file will give you lots and lots of info about your PHP installation and its capabilities.

Printer setup

Neil Youngman (ny at youngman.org.uk)
Sat Apr 1 00:01:28 PST 2006

Answered by: Ben, Martin, Neil

I upgraded to the latest SimplyMepis a while back and since then I've been unable to get it to configure my printer. IIRC, when I originally set up Mepis I just installed some extra packages to give it the right drivers and configured it in the CUPS web interface at localhost:631.

The printer is a Xerox XK35C, for which the lex5700 driver is recommended. Searching the debian packages for xk35c, I find foomatic-db and foomatic-filters-ppds seem to have relevant files. Having made sure both are installed, I still don't get the Xerox Workcentre XK35C or the Lexmark 5700 on the list of available printers.

The HOWTOs don't offer anything more than install the correct driver package and configure printer. Does anyone know what I'm missing here?

[Martin] - Have you re-started CUPS?

 /etc/init.d/cups restart 

Sometimes you have to re-start Cups for it to see any new drivers then just select the printer in the web interface...

[[Neil]] - This machine is not usually left on overnight, so it's been rebooted quite a few times since the drivers were installed.

[Ben] - Check to make sure that your system (via hotplug) still detects your printer. Examine the output of 'dmesg' for something like this:

usb 2-1: ep0 maxpacket = 8
usb 2-1: default language 0x0409
usb 2-1: new device strings: Mfr=1, Product=2, SerialNumber=3
usb 2-1: Product: psc 1310 series 
usb 2-1: Manufacturer: hp
usb 2-1: SerialNumber: CN522C5230V4
usb 2-1: uevent
usb 2-1: device is self-powered
usb 2-1: configuration #1 chosen from 1 choice
usb 2-1: adding 2-1:1.0 (config #1, interface 0)
usb 2-1:1.0: uevent
usb 2-1: adding 2-1:1.1 (config #1, interface 1)
usb 2-1:1.1: uevent
usb 2-1: adding 2-1:1.2 (config #1, interface 2)
usb 2-1:1.2: uevent
drivers/usb/core/inode.c: creating file '002'
hub 2-0:1.0: state 7 ports 2 chg 0000 evt 0002
uhci_hcd 0000:00:1d.0: suspend_rh (auto-stop)
hub 3-0:1.0: state 7 ports 2 chg 0000 evt 0000
drivers/usb/core/inode.c: creating file '001'
usblp 2-1:1.1: usb_probe_interface
usblp 2-1:1.1: usb_probe_interface - got id
drivers/usb/core/file.c: looking for a minor, starting at 0
drivers/usb/class/usblp.c: usblp0: USB Bidirectional printer dev 2 if 1 alt 0 proto 2 vid 0x03F0 pid 0x3F11
usbcore: registered new driver usblp
drivers/usb/class/usblp.c: v0.13: USB Printer Device Class driver

Also (assuming that it's a USB printer and you have 'usbfs' configured and mounted), take a look at '/proc/bus/usb/devices'; my printer entry there looks like this:

T:  Bus=02 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  3 Spd=12  MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs=  1
P:  Vendor=03f0 ProdID=3f11 Rev= 1.00
S:  Manufacturer=hp
S:  Product=psc 1310 series
S:  SerialNumber=CN522C5230V4
C:* #Ifs= 3 Cfg#= 1 Atr=c0 MxPwr=  2mA
I:  If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=cc Prot=00 Driver=(none)
E:  Ad=01(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=81(I) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=82(I) Atr=03(Int.) MxPS=   8 Ivl=10ms
I:  If#= 1 Alt= 0 #EPs= 3 Cls=07(print) Sub=01 Prot=02 Driver=usblp
E:  Ad=03(O) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=83(I) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=84(I) Atr=03(Int.) MxPS=   8 Ivl=10ms
I:  If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E:  Ad=07(O) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=87(I) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=88(I) Atr=03(Int.) MxPS=   8 Ivl=10ms
I:  If#= 2 Alt= 1 #EPs= 3 Cls=ff(vend.) Sub=d4 Prot=00 Driver=(none)
E:  Ad=07(O) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=87(I) Atr=02(Bulk) MxPS=  32 Ivl=0ms
E:  Ad=88(I) Atr=03(Int.) MxPS=   8 Ivl=10ms

If all the hardware stuff is OK, try printing directly to the device:

cat /etc/group > /dev/usb/lp0

If that works, then the problem is (obviously) CUPS. I've wrestled with that thing myself, in the past, and here's what worked for me:

Check /etc/cups/ppd to make sure that you have the correct PPD file. If you do, and 'cups' still fails to detect your printer, go to http://linuxprinting.org and grab a fresh copy of that PPD file - I've had the distro install a stale version that couldn't handle my (brand new) printer twice now, with two different printers.

[[Neil]] - No it's not USB

[Ben] - There's also a 'HOWTO'-type page at the above URL that walks you through the intricacies of dealing with CUPS and FooMatic stuff that you might find helpful.

[[Neil]] - I've looked at that before, but this time I found enough inspiration to solve it. The PPD file hadn't been installed where CUPS could see it.

[[[Ben]]] - Now that you remind me - yeah, shortly after I installed CUPS a couple of years ago, I had the same problem. The PPD I needed was located somewhere deep inside '/usr/share' or /usr/lib', but CUPS just ignored it. I think it was that FAQ at linuxprinting.org that finally put me onto the fact that it should be in '/etc/cups/ppd' instead.

[[Neil]] - I located the PPD file with

find / -xdev -iname \*xk35\*

and then I copied it to /usr/share/cups/model/

root at 2[~]# 
gunzip /usr/share/ppd/linuxprinting.org-gs-builtin/Xerox/Xerox-WorkCentre_XK35c-lex5700.ppd.gz
root at 2[~]# 
cp /usr/share/ppd/linuxprinting.org-gs-builtin/Xerox/Xerox-WorkCentre_XK35c-lex5700.ppd /usr/share/cups/model/
root at 2[~]# /etc/init.d/cupsys restart
Restarting Common Unix Printing System: cupsd.
root at 2[~]# 

After that I was able to configure it from the web interface.

Easy once I looked in the right place. Thanks for getting me on the right track.

[[[Ben]]] - Glad I could help, Neil!

Reading Oxygen Phone Manager files

Jimmy O'Regan (jimregan at o2.ie)
Thu May 25 08:17:43 PDT 2006

Answered by: Ben, Francis, Jimmy

Every now and then I use a Windows program called "Oxygen Phone Manager" to back up my phone. It's able to read information that none of the Linux programs can, so it solves the problem of how to back up everything on the phone, but uses some strange file formats of its own, which is another problem...

This script converts data from the call registry to the XML format that OPM outputs (.cll files, or [mnt point]/Program Files/Oxygen/OPM2/Data/Phones/[phone IMEI]/CallRegister.dat).

This is probably the least useful information that OPM extracts, but I needed to use it to understand the date format OPM uses, and someone somewhere might find it useful.

[Jimmy] - It turns out that at least one other program uses the same date format, so the date stuff is probably more useful than I'd thought.

Anyway, here's a Python version of the date stuff (I use python as a hex editor :)

[Francis] - Hi there,

I think there are some imperfections in the homebrew date code (which is why there are modules for that sort of thing, of course).

As a verification, 36524 days after 1 Jan 1900 should be 1 Jan 2000 (100 years x 365 days per year plus 24 leap days between those dates).

My date(1) doesn't go back as far as 1 Jan 1900, but 1000 days later is 28 Sep 1902, so I can use that as an independent verification.

   $ date -d "28 Sep 1902 + 35524 days" +%d/%m/%Y

   print days_since_1900(36524);

So: homebrew perl is a few days out, and it prints negative numbers.

   > sub split_date
   > {
   > # Number of days since 1900
   > if (eval "require Date::Manip")
   > {
   > import Date::Manip qw(DateCalc Date_Init);
   > # OPM doesn't seem to use proper leap years
   > $numdays -= 2;
   > }
   > else
   > {
   > # My crappy function, as a last resort
   > days_since_1900 ($numdays);
   > }
   > }
   > if ($buggy_as_hell)
   > {
   > # 38860.8914930556 should be 23/05/2006 21:23:45
   > split_date(38860.8914930556);

That'll give 23/05/2006 because you subtract 2 within split_date when using Date::Manip, as shown above.

38860 days after 1 Jan 1900 is actually the 25th. It's only worth mentioning because you don't explicitly subtract 2 within days_since_1900.

   > sub days_since_1900
   > {
   > my $numdays = shift;
   > my @mdays = qw(31 28 31 30 31 30 31 31 30 31 30 31);
   > my $years = int($numdays / 365);

Minor point -- after 366 leap years have passed, that'll be off.

Unlikely to be a major problem.

   > if ($buggy_as_hell) {print "$years years\n";}
   > $numdays -= ($years * 365);

That's $numdays %= 365, "the number of days gone this year plus the number of leap days since epoch".

   > if (($years % 4) == 0)
   > {
   > $mdays[1] = 29;
   > }

Other minor point -- that's strictly wrong; but for the range of "90 years either side of today" it works, so it's also unlikely to be a major problem.

   > if ($buggy_as_hell) {print "February has $mdays[1] days\n";}
   > my $leapyears = int ($years / 4);
   > if ($buggy_as_hell) {print "$leapyears leapyears\n";}
   > $leapyears++; # Um... 'cos this doesn't count 1900 as OPM does

Don't know exactly what OPM does, but $leapyears is now "the number of leap days since epoch, plus one, plus one more if this is a leap year but it's not yet Feb 29th" which is probably not a very useful count.

   > $numdays -= $leapyears; 

The big problem is here. $numdays can now be negative, which will lead to odd dates come December -- try 39050 and 39060 and spot the awkwardness.

So: if $numdays is less than 1, decrease $years by 1, increase $numdays by 365 (or 366 if the new $years corresponds to a leap year), and then think about how to handle the "if this is a leap year but it's not yet Feb 29th" thing.

Oh, and maybe subtract the 2 that happens in the Date::Manip case.

It's too much to wrap my head around right now, so I'll just suggest the cheaty

   $numdays -= 1000;
   $date=qx{date -d "28 Sep 1902 + $numdays days" +%d/%m/%Y};

as a shell-out workaround.

And the magic OPM "2"? Maybe it counts 1 Jan 1900 as day 1 not day 0, and maybe it thinks 1900 was a leap year? That could bump the current numbers up by 2 compared to what is expected, perhaps.

The same algorithmic error is in the python implementation.

[Ben] - I didn't have the time to dig into it, but two errors jumped out at me:

   import Date::Manip qw(DateCalc Date_Init);

Ummm... don't do that. Just don't. The reasons are long and complex (see 'perldoc -f import', if you feel like unraveling a tangled string), but that should be 'use Date::Manip qw(DateCalc Date_Init);' instead.

You've also got 'xml_line' being invoked as

   while (<>)
   my ($type, $order, $number, $foo, $name, @times) = split (/\t/, $_);
   xml_line ($type, $number, $name, @times);

but declared as

   my ($type, $number, $name, $stimes) = @_;
   my @times = split / /,$stimes;

Given your sample data, you're putting a single time string into '@times', then reading it as a scalar. The easiest fix (well, 'problem prevention' rather than 'fix' - it'll work as it is unless a "\t" gets into the time data) would be to invoke and process it as a scalar:

   my ($type, $order, $number, $foo, $name, $times) = split (/\t/, $_);
   xml_line ($type, $number, $name, $times); 

or, perhaps, by completely eliding those temp vars since they're only used once anyway - which would eliminate the problem as a side effect:

   while (<>)
   xml_line ( ( split /\t/)[0,2,4,5] );

Network Traffic Review/Filtering

sloopy (sloopy_m at comcast.net)
Thu Apr 6 04:27:56 PDT 2006

Answered by: Francis, Kapil, Suramya

Greets and Salutations,

I have lurked about on the ML for quite a while and wanted to thank all that make this source of info (and at times humor) for their time and knowledge.

The question(s)

I run a 8-10 node network at home through a router (a VIA C3 mobo with fedora core) and would like to have a way of setting up a web page on it that would list URL's being retrieved from the inet, and a nice side option of being able to block certain content for some nodes on the network. would i need to run a proxy (i.e. squid or similar) for this? or would this be over the capabilities of the router machine?



[Francis] - Hi there,

I have lurked about on the ML for quite a while

Great -- I'll not explicitly Cc: you since you're on the list already.

Short answer: you do not need to run a proxy, but you should do so. And it should not exceed the capabilities, unless you've other funny stuff going on.

Longer answer follows...

Usually, the hard part about designing a solution is precisely specifying the intention ;-)

that would list URL's being retrieved from the inet

will need to have access to those URLs.

I'm guessing that you primarily care about http, with perhaps an interest in https or ftp too.

Expect that you won't see details of https requests. It's easier that way.

One way to get access to those URLs would be to run a network sniffer on the router and configure it to record HTTP requests.

Completely passive, no change to what the users see or do, and you can prevent the system becoming swamped by allowing it to drop packets if it thinks it is too busy, at the cost of not having a complete record of the accesses. Google will be able to suggest software which can do this, if you don't have any favourites. One variant I've come across (but can't think of right now) involves sniffing and displaying any images in http traffic that are on the network. That may or may not be appropriate for your setup, of course.

[[Suramya]] - The software you are thinking about is called Driftnet .

Be really really careful while running this. We tried this as a test at my last job and were showing off the program to a group of co-workers and I guess someone in the building was surfing porn at that time so we got to display some really explicit content on a really big display.

Consider yourself warned if you want to use this... :)

[Francis] - Another way to get the URL list would be to require the use of a http proxy server, and read the log file it generates. One way to require the use of a proxy server is to allow (web) access from it and block access from elsewhere. Another way is to try network magic on the router and use a transparent proxy. (With the obvious caveat that a transparent proxy can't work in all cases; that may not be an issue for you.)

Small or no changes to what the users see and do. Active, in that if the extra service fails, web access is stopped. And it does allow for the possibility of caching and access control, if you go that way.

In either case, for a web page, "tail the-url-list-file" as an appropriate user might be sufficient, or just show the raw log file, or use any of the log analysers to generate pretty pictures. Again, Google will likely show you a ready-made solution for whichever you pick.

being able to block certain content for some nodes on the network.

When you have defined (for yourself) "certain content" and "some nodes", you'll probably find that it's easier to do this at the http level, with a proxy server. It can be done at the tcp level with a more active network sniffer, but the simple answer is "don't do that".

On my machine (PII 366MHz, 128MB) I use squid as a caching proxy server, with squidGuard as a redirector intended primarily to block some images and undesired content. Single client node, and (deliberately) easily bypassed. (I also have a thttpd to serve the redirected content, for completeness.)

In my setup, the control is handled mostly by squidGuard, which can decide based on source address or url, but not based on content. That's good enough for me. (There's a Fine Manual with all of the details.)

Occasionally I want to make local modifications to the block-or-allow lists, which I do from a shell. A webmin module for it exists if you like that kind of thing, but I've not examined it in a long time.

"My users" is "me", and they all know how to fix or work around broken bits without bothering their admin. This happy state may not match your network.

Extra memory is always good. And disk space if you choose to cache or retain logs. But if the machine can run Fedora Core, it can run squid/squidGuard for a small network.

Set up squid and squidGuard and whatever other bits you want, and use it as your proxy server. Configure squid to allow what you want to allow, and block the rest. Configure squidGuard to allow or block as appropriate. And make sure that "what you want to allow" won't surprise the other users.

Admire your URL list web page, and change it to match what you want.

Decide how important to you it is to block access and have a full log of http requests.

Then either invite others to use the proxy, require others explicitly to use it, or require others transparently to use it (the latter two include adjusting your ip filtering rules).

And when it's all working, if you've something to add over what's in the lg archives, write it up as an article and submit.

Good luck with it!

[[[Kapil]]] - As Suramya pointed out anecdotally, in any (re-)configuration of routers/firewalls make sure you understand and can handle the "politics".

As Francis Daly said you have three solutions. I'll add a glimpse to the politics associated with each.

a. Force all nodes to use a web proxy by blocking other nodes from accessing the web directly (using firewall rules). Any web proxy combined with a log analyzer (analog?) can do what you want.

Provide a ".pac" file (for automatic proxy configuration) for user convenience.

This way everyone using the nodes knows what you are doing and how.

b. Automatically redirect web connections from the nodes to the web proxy by firewall rules. You need a web proxy (like squid) that can handle "transparent proxying".

The users need not be told anything but they'll probably find out!

"Transparent" proxying is generally not quite transparent and in my experience does break a few (very few) sites. Note that web proxies are acounted for by the RFC for HTTP but transparent proxies are not.

c. Use firewall rules to send a copy of all web traffic through a sniffer which can extract the URL's. You can insert firewall rules to block/allow specific IP addresses.

Again the users need not be told anything.

You will not be breaking any network protocols by doing this.

Hope this helps,

Meaning of overruns & frame in ifconfig output

Ramon van Alteren (ramon at vanalteren.nl)
Thu Apr 6 13:41:50 PDT 2006

Answered by: Ben, Francis, Martin, Ramon

Hi All,

Out of pure interest would anyone know the exact definition (or provide me with pointers to said definition) of the fields overrun and frame in the output of ifconfig.

I've been searching google for an answer, but that mostly turns up false positives from all the people over the years that have posted their ifconfig output on the internet.

I've checked wikipedia which has an external link to the following short bit on overruns but nothing on the frame field: "Receiver overruns usually occur when packets come in faster than the kernel can service the last interrupt. "

I'm seeing these (overrun & frame errors) on a NIC in a load-balancer which services just the incoming http-requests (outgoing uses direct routing) and I'm buying new ones tomorrow. I am however still curious what these values actually mean.



[Francis] - The source code might :-)

[[Ramon]] - Kept that as a last resort, my C coding & reading skills is at best rusty and at worst non-existant.

[Francis] - But there's a 450 kB 45-page pdf at both

and http://www.computer-tutorials.org/ebooks/02_summer_art1.pdf

with the heading

International Journal of Digital Evidence
Summer 2002, Volume 1, Issue 2

"Error, Uncertainty, and Loss in Digital Evidence"
Eoghan Casey, MA

which includes on p29

One manufacturer provides the following information about interface errors,
including datagrams lost due to receiver overruns a.k.a.  FIFO overruns (NX Networks,
    *  packet too long or failed, frame too long: "The interface received
    a packet that is larger than the maximum size of 1518 bytes for an
    Ethernet frame."
    *  CRC error or failed, FCS (Frame Check Sequence) error: "The
    interface received a packet with a CRC error."
    *  Framing error or failed, alignment error: "The interface received
    a packet whose length in bits is not a multiple of eight."
    *  FIFO Overrun: "The Ethernet chipset is unable to store bytes in
    the local packet buffer as fast as they come off the wire."
    *  Collision in packet: "Increments when a packet collides as the
    interface attempts to receive a packet, but the local packet buffer
    is full.  This error indicates that the network has more traffic than
    the interface can handle."
    *  Buffer full warnings: "Increments each time the local packet
    buffer is full."
    *  Packet misses: "The interface attempted to receive a packet,
    but the local packet buffer is full.  This error indicates that the
    network has more traffic than the interface can handle."

[[Ramon]] - Mmm interesting link, more reading material for the stack ;-) Thanks

[Francis] - This is friend-of-a-friend third hand stuff by now, of course, but it certainly sounds reasonable to me, and might give you pointers to some more terms to search for to find something demonstrably authoritative.

For what it's worth,

/proc/net/dev packet framing error overrun

was what I was what I (eventually) asked Google for, before asking for

"NX Networks" ifconfig

which found the two links.

I hope the above isn't useless.

[[Ramon]] - Certainly not, at the very least it will serve to enlight my soul with knowledge, which is a good thing (tm)

FYI: It took me awhile to pinpoint this bottle-neck, most answers google turned up pointed to badly configured or malfunctioning hardware and or pci-busmaster stuff.

This is a well-configured Intel 100Mbit NIC with busmastering enabled which is apparently flooded with http-traffic to the point that it can not generate interrupts fast enough to get the frames out of it's buffer. That's the first time I've ever seen that.

[[[Martin]]] - DOS attack was the first thing that came to mind...

[[[Ben]]] - I've seen NICs from nominally reputable vendors start throwing a large number of errors under a much lower load than what you've described - I tested several identical units and gave up in disgust. The Intel cards that I used to replace them worked great, never a complaint. The one you described is just an amazing sample of the breed; it should be given a retirement dinner, a gold watch, and a full pension.

[[[[Ramon]]]] - It's an Intel NIC


[[[[Martin]]]] - So what would cause a network card to throw that many errors then?

I'm guessing that its either dodgy drivers or a actual dodgy manufacture...

[[[[[Ben]]]]] - The latter. This happened ~7 years ago, but since I had to send the cards to the States to get them replaced (I was in the USVIs at the time), and everything took place in S-L-O-W T-I-M-E, many of the details are still with me. It all took place under a certain legacy OS, but the drivers that the vendor specified (ne2000) were the bone-standard, came-with-the-OS types. By the time the third card arrived - I had a hard time believing that I got two bad cards, but seeing trainwrecks happening at 2MB/s was a pretty strong motivator to find out - I had installed Debian (dual-boot) on my machine, and could show the NIC falling over and twitching piteously under two different OSes. Then, I marched into the CEOs office with a couple of printouts and demanded a case of Intel NICs. I had no experience with them personally, but I had a number of trustworthy admin friends who swore by these things while swearing off everything else.

Oh yeah - I should mention that the CEO of that company was convinced that the IS department should be able to run on, oh, fifty cents a month - and asking for anything more was an outrage and a shuck that he was way too smart to fall for. This led to some interesting confrontations on a regular basis - I suppose he liked high drama. Anyway, I don't recall if I had to resort to bodily harm and talking about his mama, but I got those cards and spent the week after that installing them on every machine in the company (except the CEOs, he-heh. He had a brand-new Dell for his Minesweeper and Solitaire, and had previously told me that he didn't want me to touch it.)

After The Great Replacement, many of my network troubles disappeared - and as a bonus, so did lots of database problems. Seems that many of the latter were caused by apps establishing a lock for the duration of the transaction + the network flaking out during that transaction. This would cause the lock to persist until that machine was rebooted (!), and nobody else could get access to that DB until it was... this was a Novell 4.01 goodie that fortunately went away when I updated everything to 4.11, months later.

[[[Francis]]] - (quoting Ramon): Kept that as a last resort, my C coding & reading skills is at best rusty and at worst non-existant.

As backup for the "random file on the web" info, /usr/include/linux/netdevice.h on my 2.4-series machine includes in the definition of "struct net_device_stats", in the "detailed rx_errors:" section

  unsigned long   rx_frame_errors;        /* recv'd frame alignment error */
  unsigned long   rx_fifo_errors;         /* recv'r fifo overrun          */

which seems consistent with that information. (Similar entries appear in a few other headers there too.)

And egrep'ing some nic drivers for "rx_frame_errors|rx_fifo_errors" reveals notes like "Packet too long" for frame and "Alignment error" for fifo.

So if it's wrong, it appears consistently wrong.

As to the cause of the errors -- fifo overrun, once the nic is configured and negotiated right, might just mean "you have a (seriously) large amount of traffic on that interface". Presumably you've already considered what devices might be generating that traffic, and isolating things on their own network segment. An alternative is that the machine itself is too busy to process the interrupts and data, but you would likely have spotted that sooner.

Frame alignment, on the other hand (again, presuming that it isn't just an artifact of a nearly-full buffer) suggests that some other device is generating a dubious frame. Any consistent pattern in source MAC or IP which can be used to point the finger?

[[[[Ramon]]]] - Nope the reason I couldn't believe the error at first sight is that it is NOT failure behaviour. It simply is legit traffic ;-)

It's the external nic on the load-balancer for the website I work for. It's pulling 9M+ pageviews per day with a 12 hour sustained peak of 500K+ pageviews per hour. All that traffic needs to go through this single 100Mbit interface, in large amounts of small http request packets.

I'm swapping it out for a Gigabit card later today.

We'll probably frame the card as an achievement and hang it somewhere around the office :-D

Search Engine Spiders

bob van der Poel (bvdp at xplornet.com)
Thu Apr 20 16:46:45 PDT 2006

Answered by: Ben, BobV, Francis, Jason, Thomas

Any light on how search engines like google work? I've recently moved my web stuff to the free site provided by my new ISP. For some reason google refuses to list any of my "really good stuff". And, doing some searches I don't think that any other (or very few) pages on this site are being found. My page http://users.xplornet.com/~bvdp has a fairly unique pattern I can test "wynndel broadband xplornet". So far the only hits are back to my OLD web pages which announce a move. BTW, those pages are gone.

I've discussed this with Xplornet and they have come to the conclusion that their site is "sandboxed", but don't know why, etc. And, frankly, I'm not sure they care much either :)

I have tried to "seed" things by filling out the google form (sorry, forget the exact spot right now).

I find the whole issue pretty interesting, especially since in the past when I've created new pages, etc. they have been found by google within hours. Days at the most. These new pages have been up for about a month now and not a hint from google.

[Thomas] - Probably then, the google site is listed in 'robots.txt' on the webserver as not being valid when sending out bots to spider your site (or a subdomain). It's a common practise to ban most harvesting bots, and to allow only a few.

[[BobV]] - I've actually checked that out. The ISP assures me that there are no robots.txt files buggering things up. I should have mentioned that in my original post, I guess.

[Thomas] - Google work by sending out spiders -- programs that essentially use referral-following (URL-hopping if you like) to see what's what. The greater linked your site is, the greater the chance you'll get'spidered'. (I am sure this is explained on google's site. Go and look).

  > searches I don't think that any other (or very few) pages on this site 
  > are being found. My page http://users.xplornet.com/~bvdp has a fairly 
  > unique pattern I can test "wynndel broadband xplornet". So far the only
  > hits are back to my OLD web pages which announce a move. BTW, those 
  > pages are gone.

So? Even if your site is hit by google (or some other indexing surveyor) that's still no guarantee your search results will show up. Consier a company such as Google. Do you have any idea just how much data they house? Whenever you goto 'www.google.com', and perform a search -- you're connecting to a different cache of information each time, so the results you get returned for a known search may well differ. And note that "specific" != "likely chance of singular hit". In fact that's quite the opposite. If your site is the only one listed with a specific phrase, and has not been referenced elsewhere on the net for google to pick up, what do you think the chances of it being listed are?

  > I've discussed this with Xplornet and they have come to the conclusion 
  > that their site is "sandboxed", but don't know why, etc. And, frankly,   
  > I'm not sure they care much either :)

To me it's a moot point, and is slightly egotistical to want to see your site on a search engine. If that happens all well and good. Doubtless it might happen eventually.

[[BobV]] - Of course it is moot. Which is why I posted this message. Figured it might of some interest to others. Sorry if you think I'm just being egotistical or whatever.

[Jason] - Quoting BobV, "Any light on how search engines like google work?"

Well, you start with a novel concept and a bunch of really smart people. Then you leverage commodity hardware and open source software to create a superior product with a clean, simple design. You dominate the market, go public, and then you sell out your principles to help China suppress free speech...

...oh, you mean *technically*? No idea. :-)

[Ben] - Well, there's a surefire way to find out if Google knows about you:

ben at Fenrir:~$ google http://users.xplornet.com/~bvdp
Sorry, no information is available for the URL users.xplornet.com/~bvdp

  * If the URL is valid, try visiting that web page by clicking on the following link: users.xplornet.com/~bvdp
  * Find web pages that contain the term "users.xplornet.com/~bvdp"

I guess it doesn't. I'm assuming you went to http://www.google.com/addurl.html to add yourself - yes? And you made sure to put in a list of keywords that are relevant to your site - yes? I've always found this to work just fine for my clients, although it does take up to a couple of weeks (and it's amazing how impatient people can get in the meantime. :)

[[BobV]] - Yes, that's what I did. I also tried to set up a site map but it appears that I need permission to run python on the site for that to work. I can't run anything on the remote site, so that is out.

[[[Francis]]] - All the site map is is a list of links, yes? Probably with some organisation and grouping and background and the like, but fundamentally, a link to every page on one page? (And the hrefs shouldn't start with "http://", and probably shouldn't start with "/" either, for portability.)

Do you have a local copy of the web site content? If so, you can run a local web server configured similarly to the public one, and run the site map generating program against that. (Or use "find" with some extra scripting, but that uses a file system view rather than a web client view of the content.)

That should lead to one or a few html pages being created; upload them to the public site and you're laughing.

[[[[BobV]]]] - Yes, I think you have it pretty much correct. I've not bothered to 'dl the code ... I don't see that it does me much good to know how simple my simple little site is :) But, you idea of running in on the local site and then putting the results on my public site crossed my mind as well. Unless they have some checksums/tests in the code it should work fine.

[Ben] - So, try submitting to a bunch of other engines in the meantime. They all "steal" from each other, AFAIK - and the more places there are on the Web that refer to your site, the higher the chance that Google will run across it sooner.

[[BobV]] - You know, I get to used to using google that I forget that other engines even exist. Just for fun, I tired yahoo and, interestly, it found the site.

So, it probably is just a matter of time.

[[[Ben]]] - This should also provide a bit of reassurance on Google's score: if Yahoo indexed you, then so will they - in their own good time.

2[BobV]- Honestly, this is not a big deal with me. I do contribute some software and ideas which I suppose other folks might be interested in (which is why it's on the web). I'm just trying to understand (at a low level) how these search spiders work. I wonder, with all the computing power that something like google has ... how long would it take to "spider" the whole web? They are already referencing billions of pages and my mind gels thinking of how many links that might involve, let alone keyword combinations.

[[[Ben]]] - Oh, it's fun (and at least somewhat mind-boggling) to think in those terms. I mean, just how much space does the Wayback machine (which takes regular snapshots of just about everything on the Net) have? More interesting yet, when will that kind of capacity - and data-processing capability - become available to us consumer types?

"Say, Bob, do you happen to have the latest copy of the Net? I think I'm a bit behind; just beam me your version if you don't mind..."

4 [BobV] - Did I read somewhere, recently, that someone was marketing a disc (HD?) with "the net" on it. Well, a very small subset of the net :) I think this was for folks who didn't have broadband or something. Seemed like a silly idea to me.

So, which is growing faster ... consumer storage or the size of the net? I'd bet on the net! Mind you if we take all the porn, music, duplicate pages and warez software off the net ... well, gosh, then there would be nearly nothing left.

[[[Francis]]] - Do you have access to the web server logs? Ideally the raw logs, which you can play with yourself; but even the output of one of the analysers may be enough.

[[[[BobV]]]] - Unfortunately, no. Just a dumb site from this end.

[[[Francis]]] - If you can see the HTTP_USER_AGENT header in the logs, you'll have an idea of when the search engine spiders visit. Apart from the "normal" clients, almost all of which pretend to be some version of Mozilla, the spiders tend to have reasonably identifying names.

[[[[BobV]]]] - Interesting. No secrets on the web :)

[[[Francis]]] - If you've ever seen "googlebot" (or however it is spelled) visit, you can expect that they are aware of your content[*]. And if you've never seen "yahoobot" visit (again, I'm making up the user-agent string here -- when you see it, you'll know it) you can wonder about how they managed to index your content.

[*] Unless it's someone playing games, using that sort of identifier in their own web client. All of the HTTP headers are an indication only.

Whether the logs are available is up to the hosting company. And if they are available, whether they include HTTP_USER_AGENT or HTTP_REFERER or any other specific headers is also up to them.

But if you can get the logs, you can spend hours playing with them and finding trends and counting visitors and seeing why "unique visitors" is a fundamentally unsound concept. But there are almost certainly more enjoyable ways of spending your time :-)

Good luck

[[BobV]] - Impatient ... when you say "weeks". Gosh, Ben, these days anything more than a minute or two is an eternity :)

[[[Ben]]] - [laugh] Well, there is that. Most people have been trained to expect instant gratification these days (athough many claim that instant is too slow in the Internet Age), and a couple of weeks might as well be a lifetime. Although I understand that a baby still takes nine months, no matter how hard you work at it. It's all so confusing...

Quoting from the classic "Story of Mel",

Mel loved the RPC-4000 because he could optimize his code: that is, locate instructions on the drum so that just as one finished its job, the next would be just arriving at the "read head" and available for immediate execution.

I suspect that your previous experiences happened when your part of IP space (or whatever other system Google uses to sequence their spiders) was just about to "arrive at the read head". This is, I assure you, not the usual case. :)


Profile has errors

K1 (research.03 at gmail.com)
Thu Apr 13 16:21:06 PDT 2006

Answered by: Thomas


I am new to IPCOP and am learning it while setting it up, I wonder if you can point me to asomeone who can help me.

Your article was very informative, but my issue is with DSL.

I have Verizon DSL and am using a Westell Versalink 327W. I am utilizing the red and green interfaces.

My IPCOP box is handing out addresses as I have DHCP enabled, but I get time out errors when I try to ping it . When I access it from a browser on the green interface I get "Current profile-Non" and "the profile has errors".

When I view the log on my Westell I see my ipcop box. Help what am I doing wrong?

[Thomas] - What are these 'red and green interfaces'?

What about using ''traceroute'' instead? What's your network topology like? Are you using a gateway (perhaps some form of NAT?) Do you handle your own DNS? You need to be much more thorough in the information you detail to us.

Sounds like you need to read the IPCop userguide, or somesuch then.

You're not telling us what's in the logs, for starters. We're not mind readers -- and knowing what's in this file, might just help provide pointers as to its solution.

Where can I find GLUE (Group of Linux Users Everywhere)

Shlomi Fish (shlomif at iglu.org.il)
Sat May 6 19:56:15 PDT 2006

Answered by: Ben, Rick, Shlomi


I am the dmoz.org editor of the Linux Users' Group category. Recently, GLUE (Group of Linux Users Everywhere) has disappeared from the Net. What is its status and when will it be restored? Does it have a new URL?


Shlomi Fish

[Ben] - Well - hi, Shlomi! Pleasure to hear from you.

(Shlomi Fish is a long-time contributor to the Perl community; lately, for example, I've enjoyed his contributions on the "Perl Quiz-Of-The-Week" list - so it's a pleasure to be able to give what help I can in return, little though it may be in this case.)

A quick search of the Net seems to indicate that GLUE was being hosted by our former webhost, SSC with whom we parted ways quite a while back.

[[Shlomi]] - I see.

[Ben] - According to the Wayback Machine, though, whatever arrangement they had never amounted to much:


My best suggestion for anyone looking for a LUG is to take a look at the { http://www.linux.org/users/[LUG Registry]} - but I'd imagine that you already know that. As to GLUE, my best guess is that they're gone.

[[Shlomi]] - I see. Well, I'll try to find out at SSC.

[Rick] - Hi, Shlomi. Ma nishma? ;->

[[Shlomi]] - Hi. Ani Beseder.

[Rick] - GLUE seems to have been completely eliminated in the most recent Web site reorganisation at SSC. When that happened, because I reference it in the Linux User Group HOWTO, I looked in vain for any sort of explanation, so I suppose none will be forthcoming.

[[Shlomi]] - OK. Thanks. I'll try to find out more at SSC.

Graham Jenkins

Derek (derek.davies at telco4u.net)
Sun Apr 30 11:46:14 PDT 2006

Answered by: Ben, Thomas

Dear Gang, Can you help me ?,my old school friend Graham Jenkins moved to Australia many years ago ,is he your Graham Jenkins?.

My friend was born in Hereford England on the third of April 1940,attended Lord Scudamores boys school has three sisters.

This may be a shot in the dark,but I would like to see the old fellow again before I pop my clogs.

I have been trying to find him for quite a while.

hope you can help

Derek Davies

[Thomas] - You could always look at his author page, and email him. IIRC he might be claiming to be from Australia.

[[Ben]] - That is indeed what his bio says; "a Unix Specialist at IBM Global Services, Australia... lives in Melbourne".


  > My friend was born in Hereford England on the third of April 1940,attended
  > Lord Scudamores boys school has three sisters.

[[Ben]] - The picture in his bio looks a bit younger than that, but it's plausible.

[Thomas] - Scudamores' school, eh? I know of it through reputation.

  > This may be a shot in the dark,but I would like to see the old fellow again
  > before I pop my clogs.

[Thomas] - That sounds like an awfully odd thing to say.

[[Ben]] - Google shows 981 usages. :)

[Thomas] - I must admit that your question is singularly unique. ;) I don't think Jenkins is a member of TAG. If you'll permit us to do so, we'll add this to the Gazette Matters section of LG.

[[Ben]] - I note (again, via Google) that Graham is quite active in the Perl community and elsewhere. Perhaps emailing him at the address on his page would be the most efficient way of getting hold of him.

Request for any ADC's driver source code

osas e (osas53 at yahoo.com)
Sat May 20 10:43:13 PDT 2006

Answered by: Thomas


I am a student doing a project and want to interface an ADC0804 to a PC’s parallel port to read the humidity value of the atmosphere. Knowing that there are professionals like you out there, I am writing to request a driver source code to read from this or any other ADC using a PERL Program. It is my belief that you will be kind enough to help a newbie like me grow in computer interfacing.

Kind Regards

Ediae Osagie

[Thomas] - Tell us more about this project. Does your project also stipulate that you yourself have to work out how to interface an ADC0804 to the paralell port?

As a "professional" (and a student) it is usually a given that a student does their homework beforehand. How is it you heard of us, anyway? You can't have read the Linux Gazette, since you'd know that we don't do others' homework. You're no different from this.

Having done some electronics in the past (these are just pointers for you to consider), interfacing to the paralell port is relatively easy -- assuming you do so in an appropriate language. C for instance has many different routines to achieve this, and in fact, any simple I/O system can read and write to /dev/lp* as is necessary. Try it. That's the only "driver" you'll really need.

Enjoy your homework -- we won't help you do it.

Talkback: Discuss this article with The Answer Gang

Published in Issue 128 of Linux Gazette, July 2006

The LG Talkbacks


Talkback:124/smith.html (7)
Talkback:124/smith.html (8)
Talkback:124/smith.html (9)
Talkback:126/cherian.html (1)
Talkback:126/cherian.html (2)
Talkback:126/cherian.html (3)


[ In reference to the article Stepper motor driver for your Linux Computer in LG#122 ]

(jovliegen at gmail.com) jovliegen at gmail.com
Sun Apr 16 05:20:36 PDT 2006

Answered by: Ben, Jason

Hello, First of all, I like to thank you guys for this great "Gazette". I realy do enjoy it !!!

I'm trying to build the stepper motor driver, that you published in issue 122. While reading the code, I have a little remark. I'm not quite sure that I'm right, but ... what do I have to loose :D

On line 20 : "static int pattern[2][8][8] = {"

Shouldn't that be pattern[2][2][8] ?

Like I said before, I'm not sure of this ... just wondering ...

Thanks again for your nice work, all of you

[Jason] - Thanks! It's great to hear from readers.

I went and looked at that author's code, and it sure looks like you're right. Note that it doesn't hurt anything to declare the array as larger than you need it, but it's probably not the best style.

There's a few other things in that code that I might have done differently. For instance, his "step" function looks like this:

int step()
        if(k<8) {
//              if(pattern[i][j][k]==0) {
//                      k=0; 
//                      printk("%d\n",pattern[i][j][k]);
//                      k++;
//              }
//              else {
//      }	
        else  {
                printk("%d\n",pattern[i][j][k]); /*#####*/
                k++; /*#####*/
        return 0;

Note that this does not have the exact same behavior as the original function. In the original function, after step() returns, k is in range [1,8], inclusive. In the version I give, k is kept in the range [0,7], inclusive, which is the correct range if k is being used as the index of an array of length 8.

Unless I'm really missing something, I don't think the actual code to write the parallel port was included in that article.

Thanks for taking the time to point this out.

[[Ben]] - As I recall, the author used a shell script to prod the device.


[ In reference to the article uClinux on Blackfin BF533 STAMP - A DSP Linux Port in LG#123 ]

Robin Getz (rgetz at blackfin.uclinux.org)
Sat May 6 09:09:25 PDT 2006

Answered by:

In http://linuxgazette.net/123/jesslyn.html Jesslyn Abdul Salam wrote:

   >The Blackfin processor does not have an MMU, and does not provide any 
   >memory protection for programs. This can be demonstrated with a simple program:

This is only true if you leave this feature turned off.

The Blackfin processor does include hardware memory protection, which the uClinux kernel supports - but since it effects performance, and most embedded developers only turn this on when doing development, and it is turned off by default.


When you take:

   int main ()
            int *i;
            printf("%i : %x\n", i, *i);

and run it on the Blackfin/uClinux, you get:

   root:~> uname -a
   Linux blackfin #4 Sat May 6 11:38:53 EDT 2006 
   blackfin unknown
   root:~> ./test

Which is pretty close to my desktop...

   rgetz at test:~/test> uname -a
   Linux test 2.6.8-24.18-smp #1 SMP Fri Aug 19 11:56:28 UTC 2005 i686 i686 
   i386 GNU/Linux
   rgetz at test:~/test> ./test
   Segmentation fault


[ In reference to the article Interfacing with the ISA Bus in LG#124 ]

Abhishek Dutta (thelinuxmaniac at gmail.com)
Mon May 8 01:11:49 PDT 2006

Answered by: Ben, Thomas


Could you please change address my website in the article http://linuxgazette.net/124/dutta.html published in MArch,2006 issue of linuxgazette.net

You can get more details and photos related to this project at http://www.myjavaserver.com/~thelinuxmaniac/isa

The website address has changed from mycgiserver.com to myjavaserver.com


[Thomas] - I wouldn't have thought so. Most of the mirrors have probably synched by now, not to mention copious other sources. By changing your article now, that's be known as a fork -- something not desirable at all.

[[Ben]] - Even more importantly - since 'creating a fork' for already-published content doesn't bother me much - it would be essentially pointless; we currently have nearly 60 mirrors and several translation sites, and the only site that we can change is ours (i.e., the root site.) In other words, even if we change it - which I'll do, since it's trivial - anyone searching for your article will still be about 70 times more likely to find the old address rather than the new one.

For anyone contemplating writing an article, this is why I encourage including the pertinent images, sources, etc. as part of the submission: external links die. The author's details - including the URL in their bio - are flexible, since the bios are reloaded monthly by all the mirrors; the article content, once published, is essentially frozen in time.

[Thomas] - The best you can hope for is that I/we/someone publishes this request in the gazette for next month. ;) Of course, given that the talkback feature points here, and not to your personal address, you can at least be assured that any correspondence will reach us and not some void.

[[Ben]] - Excellent point, Thomas! I hadn't considered that particular benefit of it.

Talkback:124/smith.html (7)

[ In reference to the article Build a Six-headed, Six-user Linux System in LG#124 ]

Eric Pritchett (eric at gomonarch.com)
Mon Apr 24 15:07:31 PDT 2006

Answered by: Ben, Jimmy, Karl-Heinz

How would you configure this setup to use usb speakers for each user. I think the ideal solution would be to have a usb keyboard with a built in usb hub that has two usb slots, so you can plug in your mouse and usb speakers. Any thoughts?

[Jimmy] - That would solve the simpler problem: that of having the speakers close to the user; it wouldn't address the other difficulties of using multiple sound cards: of ensuring that the intended sound is directed to the correct sound card, in particular.

Ob. mailing list requests: please don't send HTML email (there's a not-so-tenuous connection between postal workers and the frequenters of mailing lists: you have been warned :), and should you reply, please don't top post: top posting makes the Baby {Jesus, Buddha, Xenu, etc.} cry. :)

If you understood my first paragraph, feel free to stop reading, as (bearing in mind that e-mail to this list is intended for publication) there is a fairly common misconception here; otherwise: USB is smart, but not that smart - the fact that devices are located close to each other physically is not communicated to the OS (that is, the system can't tell that soundcard A is connected to keyboard A, soundcard B to keyboard B, etc.), which is therefore unable to act accordingly.

[[Karl-Heinz]] - If the keyboard is its own hub -- the usb tree hierarchy should be quite able to mirror the physical proximity. I've not read the beginning of this thread -- but I can imagine a setup with two heads and two keyboards where "usb speakers" are dedicated to sound produced on one of the heads.

Just don't expect a standard distribution to do this for you -- this is manual configuration and it can not be saved in the users settings (like kde) if the users could be on either head.... pam extensions like group setting on login for devices like audio might be able help with the setup.

[[[Ben]]] - Interesting idea, that. I can imagine a system - something like LDAP in general scope if not in implementation - that assigns configured user resources from the available pool... hm, wish I knew a bit more about creating a project like that. It wouldn't even have to involve kernel programming, since everything could be done at the USB level.

Talkback:124/smith.html (8)

[ In reference to the article Build a Six-headed, Six-user Linux System in LG#124 ]

Claude Ferron (cferron at gmail.com)
Thu Apr 27 05:30:25 PDT 2006

Answered by: BobS

any update on the bugs you were experiencing?

[BobS] - Not really. There is a new nVidia driver which does not fix the problem. Two members of the Xephyr project have written to say that the problem I'm seeing is due to a reset problem in the nVidia driver. Their solution is to run several X sessions inside one big X session. This way the boards only get reset once. The URL for Xephyr is at: http://www.c3sl.ufpr.br/multiterminal/index-en.php .

Talkback:124/smith.html (9)

[ In reference to the article Build a Six-headed, Six-user Linux System in LG#124 ]

Izzy esteron (izzyesteronjr at yahoo.com)
Sun May 21 06:58:01 PDT 2006

Answered by: Ben, Izzy

To whom it may concern,

This article by Mr. Smith is amazing. In my opinion, this is the solution all 3rd world countries are looking for to keep hardware cost down. I know the technology is not perfect yet, but it is a great start.

Anyway, does this setup allow each workstation to have their own audio? How about network/LAN gaming?

Thank you for your time,


[Ben] - Well, the former was noted as a problem in the article; there has also been some discussion on the issue. What it looks like to me is that it would be possible to make this available with a little work in the kernel - but it would either take someone experienced in audio and kernel hacking, or convincing the current maintainers that this is necessary. Conversely, individual external (USB) sound devices might be a solution.

The latter does not seem like it would be a problem at all as long as the CPU and the memory is up to it; most of the load in gaming, as I understand it, is on the video card - and every user has one of their own in that configuration.

[[Izzy]] - Thanks for the reply, it was really helpful.

Last February Mr. Smith posted a message on http://blog.chris.tylers.info . He mentioned that

  "The system is VERY unstable. I get a kernel oops
  fairly often when a user logs out. Has anyone seen
  this problem before? Any ideas on how to fix it? Also,
  any suggestions to improve my article?
  Bob Smith"

Also, there was a couple of replies after this message and showed how to solve the problem by downgrading the nvidia drivers. Has Mr. Smith tried the "fixes"?

Thanks again.


[ In reference to the article HelpDex in LG#125 ]

Diego Roversi (diego.roversi at gmail.com)
Tue Apr 4 01:49:33 PDT 2006

Answered by: Ben, Rick, Suramya

I can see the comics. What's wrong with the old good image formats? Why not PNG or JPG?

[Rick] - Guys, plainly we need to FAQ this.

Diego, this was discussed in the prior issue (#124). Please see: http://linuxgazette.net/124/misc/nottag/flash.html

Note open-source (reverse-engineered) Flash implementations in various degrees of completion, detailed at http://osflash.org/ . Note also excellent and authoritative analysis by Kapil Hari Paranjape at the first URL cited. (Kapil views HelpDex using the open-source swfplayer software, which he finds adequate to the task. I recommend that plus the Flashblock browser extension, http://flashblock.mozdev.org/, so that you retain the ability to view Flash only when you wish, not when some advertiser so dictates.)

Diego, you're welcome to work with us every month to convert the Ecol comic strip to formats you prefer -- without the results sucking, please -- or become a cartoonist and start sending us good cartoons in a graphics format you prefer. Otherwise, your answer would seem to be as above.

[[Ben]] - I don't know that an FAQ entry would be of much use; in my estimate, the percentage of our readers who have looked at our FAQ is very small.

[[[Rick]]] - Interestingly, one of the less-appreciated but substantial benefits of a FAQ has nothing to do with whether people ever bother to read it before asking its questions or re-raising its tired old subjects. That is: When somebody does go there for the 1000th time, you can definitively answer, with near-zero effort, with just a URL.

[[[[Ben]]]] - Which presumes remembering the URL in question... but I do see your point. That "secondary" use of the FAQ has even more benefits than I thought - all natural consequences, to be sure, but my mind wasn't quite twisty enough to see all the rami. In short, kudos.

[[[[[Rick]]]]] - Why, thank you, sir.

You might enjoy the slightly cranky bit of documentation that I posted to the prototype wiki replacement for Silicon Valley Linux User Group's Web site (on one of Heather's machines), starting with header "Standards, Nits, Peeves" on this page:


I was definitely getting in touch with the curmudgeon within, when I wrote this entry:

   website/Website: No such word. Correct to "Web site".

   No, there's nothing wrong with inventing new words, especially where
   they make our language more expressive and fill a need not met by
   existing words. This one doesn't qualify, and is really just the result
   of people being sloppy. Some would call our cleaning it and similar
   things off our pages "prescriptivism"; I prefer the term "leadership".

[[[Rick]]] - Over the years, I've used this trick to dispose of innumerable topics after getting tired of discussing them. My first-level expectation is that hardly anyone will read my answers until I send their URLs. Note: I do take advantage of that expectation to expose readers to adjoining text on related questions.

[[[[Ben]]]] - Sneaky. And useful.

[[Ben]] - However, it might make sense to add some sort of a note regarding this issue to the top of the strip. I'll give the wording some thought and glue it in.

[[[Rick]]] - My opinion, yours with a small fee and disclaimer of reverse-engineering rights: Less is better. E.g.,

'' Format: Flash ''

...where "Flash" is anchor text hyperlinking to a FAQ, for which http://linuxgazette.net/124/misc/nottag/flash.html might serve nicely.

Metacomment: Computerists tend to overestimate public willingness to deal with explanatory text. The public at large tends to react badly to verbosity, assuming its contents to be

  • unimportant, and/or
  • open to debate

This is why we have "STOP" signs, rather than "Stop, though of course there are lots of exceptions including the need to do anything required to avoid dangerous situations" signs.

[[[[Ben]]]] - [laugh] Kat just finished her Power Squadron course (scored 100% on her test, woo-hoo!); this is almost exactly the phrasing used for right-of- way rules in navigation. Which is why, I suspect, so many people screw up in that regard.

[[Suramya]] - Umm... I have a low tech approach to converting the Ecol comic strips from flash:

Click on the link to view the flash file, take a screen shot and save it in the format you want. Sample output:



If I get the files early enough I volunteer to do this. But let me know what you think of this...

[[[Ben]]] - 1) At the moment, there's no way to do actual thumbnails linked to full-size images in the cartoon section. Changing that would require twiddling Mike's Python scripts - and I'd have no idea how to do that.

2) JPG files are significantly larger than the equivalent SWFs - and go all jaggy when viewed at anything but "native" size (unless, of course, you go with an even larger file size.) E.g.:

:-r!ls -l /tmp/Help*
-rw-r--r-- 1 ben root 80715 2006-04-06 12:50 /tmp/HelpDex.jpg
-rw-rw-rw- 1 ben root 18546 2005-12-06 21:21 /tmp/HelpDex.swf

Add a thumbnail to that, and we've got a big chunk'o'download - for no gain (and, in fact, image quality loss.)

3) The issue is - except for the very rare complaint - settled. There's Open Source viewing software available, which seems to work fine, and those who complain will now be redirected (either from here or from the link on the page itself) to the discussion covering it.

So - thank you for the offer, Suramya, but there's no need. It's all good to go.


[ In reference to the article A Brief Introduction to IP Cop in LG#125 ]

André Fellows (andrefellows at hotmail.com)
Tue Apr 11 13:43:21 PDT 2006

Answered by: Ben

Hi folks!
This IP Cop is another leaf of the Linux Firewall/Router/DHCP server distros. I have to tell you that this IP Cop seems to me not good as BrazilFW.

I used Coyote (http://www.coyotelinux.com/ ) but it was discontinued. The BrazilFW (www.brazilfw.com.br) is the Coyote continuation.

Try there, its free, fast and painless!



[Ben] - Then the obvious thing to do is a bunch of research comparing the two and presenting your results in an article. Otherwise, thank you for telling us how it "seems" to you... but I'm not sure why you'd want to share this information. [1]

I've used Coyote in the past on several occasions, and liked it a lot; for basic routing and firewalling, it was quite a nice gadget. However, IPCop has many more features than Coyote did - at least so I gathered from the article.

DIfferent pieces of software, different purposes. One is not "better" than the other - just different. Implying otherwise, especially without supporting data, is rather inconsiderate toward the person/people who took the time to write the software that you don't like.

[1] This is a (perhaps somewhat testy) way of pointing out that saying "program $X is good" does NOT require adding "because program $Y sucks" for the purpose of emphasizing the former; there's no subordinate clause implied. Since this is not the world of proprietary software, in which vendors are fighting tooth and nail to sell their latest crap-a-riffic bugware, *we do not have to denigrate competing software* - it brings us, as a community, no benefit, and may do damage to a fellow programmer's reputation. So, yeah, this hits the rant button for me.

[[André]] - My sincere excuses!!!

I forgot to mention that my "seems" is not to consider!

[[[Ben]]] - Relax - my mini-rant was a generalized one. I replied to your statement, but it's about more than that; it's about an important consideration for the Linux community in general. We should always be striving to support each other, not tear each other down.

Often, operating in the world of Open Source requires re-thinking the "standard" ways of interacting with the other people involved. This differs from what we're generally used to, but is usually easy to adapt to: substituting cooperation for competition, especially when the benefits are obvious and immediate, presents little difficulty for most people.

It does, however, require thinking about it ahead of time - which is what I'm trying to get people to do with my rant. Thanks for the opportunity. :)

[[André]] - My intention was to tell you guys about the BrazilFW, the Coyote sucessor. Like Coyote, BrazilFW have many addons to improve its funcionality. The "seems" part was because I thinked you guys could make an article comparing both and its funcionalities.

[[[Ben]]] - Well, we don't generally write articles - but we do accept well-written ones on Linux-related topics. If you'd like to write one about BrazilFW, I'd be happy to look at it.

[[André]] - This is the point of the information. Really sorry if I wasn't clear (and I really wasn't ).

Very very very sorry to make you Benjamin spend your time on mine nonsense comparison...

[[[Ben]]] - Don't worry - there was nothing particularly awful in your email, and I wasn't upset.

[[André]] - Oh, by the way, Linuxgazette is a GREAT SITE with GREAT information!!!

[[[Ben]]] - Thank you, André! We do our best. Glad to hear that you're finding it useful.

Talkback:126/cherian.html (1)

[ In reference to the article Preventing DDoS attacks in LG#126 ]

Thomas Adam (thomas at edulinux.homeunix.org)
Tue May 2 13:49:25 PDT 2006

Answered by: Ben


I was intrigued by this article about DDoS. It's something one hears about more and more, and so I thought I'd give it a go, and play along at home. Ben, I don't know just what percentage of the original article remains, having done the rewrite on it (something that I agree with, especially in terms of attribution), but I would have liked it if the original author could have expanded on a few things...

  *  Check if your CPU load is high and you a have large number of HTTP 
     process running

Heh. Yes, I quite agree. But so what? I mean, how are you supposed to distinguish that from most other processing that goes on? A simple 'w' isn't going to help you much there. It's pretty normal for most CPUs to peak.

As for the HTTP process, this is typically mitigated by setting the following in /etc/apache/httpd.conf to a reasonable value:

  MaxClients 30 

Now, the Apache folks make it quite clear that this should not be set too low, blah, blah. But for a small-end server (perhaps even one running on a home ADSL-connection) setting this to something perhaps lower than 30 in this case would help in stopping Apache from consuming everything.

Then there's the following command:

  ps -aux|grep -i HTTP|wc -l   

I know I can be pedantic at times, but on non-BSD systems the leading '-' causes issues, not to mention the classic "You'll return grep too" syndrome, as well as the useless use of 'wc' (this isn't for your benefit, Ben, more the original author):

  ps aux | grep -ic '[h]ttp'
  *  Determine the attacking network 

This section was quite good. Note that, as mentioned before, the number of concurrent connections via Apache can be capped. Again, the command:

  netstat -lpn|grep :80|awk '{print $5}'|sort 

Can be reduced to:

  netstat -lpn | awk '/:80/ {print $5}' | sort 

(Too many people seem to think Awk only understands columns in terms of quoting printed ouput. Such a shame.)

It's worth mentioning here the use of PAM and chroots. A slight side-note is that in any DDoS attack, the targeted process or thread is generally going to expand as the load increases on it. This is generally undesirable, and so one can tell PAM to employ memory usage limits via /etc/security/limits.conf (think ulimit for processes). As for the chroots, running various processes inside those can encapsulate their environment away from everything else.

I also couldn't see what the following code was supposed to do:

  for f in /proc/sys/net/ipv4/conf/*/rp_filter  
          echo 1 > done  
          echo 1 > /proc/sys/net/ipv4/tcp_syncookies  

I assume that's meant to read:

  for f in /proc/sys/net/ipv4/conf/*/rp_filter  
          echo 1 > "$f" 

  echo 1 > /proc/sys/net/ipv4/tcp_syncookies 

Hmm. This reads as though I've picked the hell out of the article for no apparent reason -- far from it. I hope these little tidbits of information will prove useful. By all means feel free to forward these comments on to the other author.

[Ben] -

On Tue, May 02, 2006 at 09:49:25PM +0100, Thomas Adam wrote:
>  Hello,
>  I was intrigued by this article about DDoS. It's something one hears about
>  more and more, and so I thought I'd give it a go, and play along at home.
>  Ben, I don't know just what percentage of the original article remains,
>  having done the rewrite on it (something that I agree with, especially in
>  terms of attribution), but I would have liked it if the original author
>  could have expanded on a few things...
>  *   Check if your CPU load is high and you a have large number of HTTP
>      process running 

> Heh.  Yes, I quite agree.  But so what?  I mean, how are you supposed to
>  distinguish that from most other processing that goes on?  A simple 'w'
>  isn't going to help you much there.  It's pretty normal for most CPUs to
>  peak.

I'm a fan of iostat/vmstat, myself - but 'w' will give you the standard 1/5/15 load averages.

That's not much to do with peaks. (Just FYI, I avoided changing the actual content of the article as much as possible except where it was clearly wrong. Different usage, or even suboptimal usage, I left alone. Ditto the idiom, which is why it still sounds mostly like Blessen rather than me.)

 As for the HTTP process, this is typically mitigated by setting the
>  following in /etc/apache/httpd.conf to a reasonable value:
>  MaxClients 30
>  Now, the Apache folks make it quite clear that this should not be set too
>  low, blah, blah.  But for a small-end server (perhaps even one running on
>  a home ADSL-connection) setting this to something perhaps lower than 30 in
>  this case would help in stopping Apache from consuming everything.

Right, but he's talking about a server in a data center - I believe that was mentioned right at the beginning.

>  Then there's the following command:
>  ps -aux|grep -i HTTP|wc -l
>  I know I can be pedantic at times, but on non-BSD systems the leading '-'
>  causes issues, not to mention the classic "You'll return grep too"
>  syndrome, as well as the useless use of 'wc'  (this isn't for your
>  benefit, Ben, more the original author):
>  ps aux | grep -ic '[h]ttp' 

[Nod] Agreed.

>  It's worth mentioning here the use of PAM and chroots.  A slight
>  side-note is that in any DDoS attack, the targeted process or thread is
>  generally going to expand as the load increases on it.   This is generally
>  undesirable, and so one can tell PAM to employ memory usage limits via
>  /etc/security/limits.conf  (think ulimit for processes).  As for the
>  chroots, running various processes inside those can encapsulate their
>  environment away from everything else.

Again, chrooting is something I'm a big fan of. Heather told me about it ages back, and I glommed onto the idea like a drowning man onto a log...

>  I also couldn't see what the following code was supposed to do:
>  >  for f in /proc/sys/net/ipv4/conf/*/rp_filter 
>  do
>  	echo 1 >  done 
>  	echo 1 >  /proc/sys/net/ipv4/tcp_syncookies 
>  I assume that's meant to read:
>  for f in /proc/sys/net/ipv4/conf/*/rp_filter 
>  do
>  	echo 1 >  "$f"
>  done
>  echo 1 >  /proc/sys/net/ipv4/tcp_syncookies 
>  Hmm.  This reads as though I've picked the hell out of the article for no
>  apparent reason -- far from it.  I hope these little tidbits of
>  information will prove useful.  By all means feel free to forward these
>  comments on to the other author.

Yeah, I missed that in the welter of the other stuff I'd corrected. Well spotted! I'll pass it on - and it should go into the Mailbag as well, as a followup to the article.

Talkback:126/cherian.html (2)

[ In reference to the article Preventing DDoS attacks in LG#126 ]

Ulrich Alpers (ulrich.alpers at ub.uni-stuttgart.de)
Thu May 4 03:00:01 PDT 2006

Answered by: Ben, Thomas, Ulrich


if I am the 10000th to mention, sorry for that ...

The last script snippet looks a bit weird:

   for f in /proc/sys/net/ipv4/conf/*/rp_filter 
   	echo 1 > done 
   	echo 1 > /proc/sys/net/ipv4/tcp_syncookies 

How about this:

   for f in /proc/sys/net/ipv4/conf/*/rp_filter
        echo 1 > $f
   echo 1 > /proc/sys/net/ipv4/tcp_syncookies

As to the syncookies thing: I am not quite sure, but if you have already put the line

   net.ipv4.tcp_syncookies = 1

into the sysctl.conf - what is the syncookies line in the rc.local for?


[Ben] - Come to think of it, it's only a couple of days after publication - and the error is egregious enough that it shouldn't be propagated. [sounds of axes, sawing, hammering, copying, and pasting [1] in the background] All fixed now.

How about this instead:

   for f in /proc/sys/net/ipv4/{conf/*/rp_filter,tcp_syncookies}
           echo 1 > $f

Love those curly-brace Bash expansions. :)

[[Thomas]] - Horses for courses -- both ways'll work.

[Ben] -

>  As to the syncookies thing: I am not quite sure, but if you have 
>  already put the line
>     net.ipv4.tcp_syncookies = 1
>  into the sysctl.conf - what is the syncookies line in the rc.local 
>  for?

Right; that should have been (and now is) worded as

  Conversely, you could add this code to your '/etc/rc.local':  

[[Thomas]] - But this is still wrong. /etc/rc.local is not honoured across all Linux distributions. SuSE and RH use it (although SuSE for a time used to only honour /etc/local.rc). RH uses /etc/rc.boot. Debian, for instance, has /etc/bootmisc.sh, but one shouldn't go changing that arbitrarily anyway (it'll get overwritten when 'initscripts' is updated). For the Debian people, one can create the following directory:


... and place any scripts in there, ensuring that the directory and
scripts themselves have octal permissions '755'.

[[[Ben]]] - That would be the reason for the word "conversely", so I wouldn't call it "wrong" - just different. Is there a distro out there that doesn't have an "/etc/sysctl.conf"? I thought that was pretty much universal in Linux.

On the gripping hand, there's also the fact that we don't purport to cover all distros - although I like to encourage authors to broaden those "$DISTRO_OF_CHOICE-only" articles into something a bit more useful. The way I figure it is, mentioning a couple of different methods should suffice to clue people in about adapting it to their specifics. If you have a better suggestion that will, without fail, cover any and every distro (including the one that I'm creating right now for the specific purpose of evading whatever solution you come up with :), I'd like to hear it.

[Ulrich] - SuSE uses /etc/init.d/boot.local (among a lot of other boot.* files).

BTW, as we are collecting the locations of the script to put the code into - Gentoo uses /etc/conf.d/local.start

[Ben] - Between Thomas Adam (who first noted the problem) and yourself, you two have managed to get me off my lazy butt and fix it. Consider yourselves promoted, and entitled to wear The Grand and Sublime Order Of The Linux Gazette [2]. Thank you, congratulations, and may you remain worthy of this high honor! :)

[1] For those who are not familiar with the BOFH Saga, the sound of these last two operations is precisely like the sound that will
immediately precede the End of The World:


[2] This looks just like a red stick-on Post-It dot, but is imbued with special powers; for this week only, we've secretly replaced the Post-It dots in your local stores with these GaSOOTLGs, which will be activated as soon as you buy and apply them.

Talkback:126/cherian.html (3)

[ In reference to the article Preventing DDoS attacks in LG#126 ]

René Pfeiffer (lynx at luchs.at)
Sun May 7 15:01:20 PDT 2006

Answered by: Ben

Hello, Linuxgazette!

I read Ben's introduction to the "Preventing DDoS attacks" article. I can imagine that you get lots of submissions like this. I've given the idea of having stand-by co-authors some thought. I like the suggestion very much, especially if it helps to "rescue" content which could interest a wide audience. However this is clearly much more than proofreading. It involves communication with the author, and it requires a change of style, obviously. I think it is a challenge to walk the fine line between helping someone and completely changing the original idea into something very different.

I am teaching at a technical academy here in Vienna. I supervised and helped several students working on their graduation. Last year a student submitted a piece of work that we could not accept, even with all the best intentions. We gave the student a long list of improvements and had him rewrite the whole thing from scratch. Of course this is a different approach, because no one can assume co-authorship for a diploma our students are supposed to write alone. Nevertheless it illustrates the point of enabling someone else to improve something.

Provided you have a pool of stand-by co-authors and co-authorship is welcome, it would be nice to have a simple set of rules how to proceed in such a case. I don't think that it is a good idea to advertise this as a feature and to encourage people to send half-written articles. Maybe postponing an article with a list of suggestions to the author is an option then.

Well, that's my brainstorming. English is not my native language, but in case you want to create a pool of helping hands that co-author and improve articles, count me in.

Best wishes, René.

[Ben] - [Quoting René]

  > Hello, Linuxgazette!

Hi, René! Good to hear from you again.

    > I read Ben's introduction to the "Preventing DDoS attacks" article. I
    > can imagine that you get lots of submissions like this.

These days, it's averaging a bit more than one per month.

    > I've given the
    > idea of having stand-by co-authors some thought. I like the suggestion
    > very much, especially if it helps to "rescue" content which could
    > interest a wide audience.

The original suggestion came from my wife, Kat, some months ago. Ever since then, I'd been planning to write a Back Page (i.e., get up on my soapbox), propose the idea, and invite participation - but somehow never got around to it. Except, of course, now - in a fit of desperation.

    > However this is clearly much more than
    > proofreading. It involves communication with the author, and it requires
    > a change of style, obviously. I think it is a challenge to walk the fine
    > line between helping someone and completely changing the original idea
    > into something very different.

Indeed. Many of these articles express an important viewpoint that's just poorly-enough stated, or address an interesting technical point but fail technical review by an amount just beyond trivial. If it's a case of pointing out the errors to the author and having them do a rewrite, or perhaps directing them to try restating their point a bit better, fine - as I keep telling people, "there's always the next issue". But what if the problem is caused by the author lacking just that tiny bit of knowledge, or that edge of ability to express themselves well in what may be a foreign languge? I hate to turn those down because I can see many of these authors trying their best, and missing by just ->that<- much. If we can find some volunteers, this would be a great resource.

    > I am teaching at a technical academy here in Vienna. I supervised and
    > helped several students working on their graduation. Last year a student
    > submitted a piece of work that we could not accept, even with all the
    > best intentions. We gave the student a long list of improvements and had
    > him rewrite the whole thing from scratch.

I almost never simply turn an article down; if I reject one, I'll have a long list of suggestions and examples that an author could use to improve their article and learn to write better articles overall. This is, as I see it, one of the duties of an editor.

    > Of course this is a different
    > approach, because no one can assume co-authorship for a diploma our
    > students are supposed to write alone.

What??? "Alone", as in "by themselves"? Shocking. If it wasn't for those famous co-authors, Messieurs Ibid, Opcit, and Anon, I suspect that fully 90% of the modern "original contributions" would never have seen the light of day. :)

    > Nevertheless it illustrates the
    > point of enabling someone else to improve something.
    > Provided you have a pool of stand-by co-authors and co-authorship is
    > welcome, it would be nice to have a simple set of rules how to proceed
    > in such a case. I don't think that it is a good idea to advertise this
    > as a feature and to encourage people to send half-written articles.

[Nod] All joking aside, those are wise suggestions. I think that half-baked articles have a substantially different quality from the ones that I find to be this sort of a moral dillemma, and I have no problem bouncing those back to the author - but, yes, co-authors would be a resource to be assigned with care and forethought.

    > Well, that's my brainstorming. English is not my native language, but in
    > case you want to create a pool of helping hands that co-author and
    > improve articles, count me in.

René, I'll take your diction and clarity of expression over many a native English speaker I've known - no joke. Your offer is gladly accepted. I don't know that I'll have anyone to send to you anytime soon (I'll be filtering submissions very carefully in this regard), but I'll definitely keep this possibility in mind. Thank you!



[ In reference to the article From Assembler to COBOL with the Aid of Open Source in LG#126 ]

(trevor at haven.demon.co.uk)
Tue May 2 02:19:42 PDT 2006

Answered by: Ben, Edgar, Jimmy, Trevor

Tag guys 'n' gals,

Converting 6000 lines of old assembly code to, err, cobol ? Why ? I would estimate that since a simple line of say c#/c++/pascal/basic compiles down to at least 6 lines of assembly and probably many more then we are looking at less than 1000 lines of 'high' level code. So it strikes me it would be quicker and less bug prone to just re-implement the program in a modern language and gain a few advantages along the way readability included.

It would have been nice if the original code/cobol code had been linked to or a summary or explaination of the programs function was included.....


[Ben] - [quoting Trev] "Converting 6000 lines of old assembly code to, err, cobol ? Why ? "

I'd imagine that it's due to The Golden Rule: He Who Has The Gold, Makes The Rules.

> I would estimate that since a simple line of say c#/c++/pascal/basic 
> compiles down to at least 6 lines of assembly and probably many more 
> then we are looking at less than 1000 lines of 'high' level code. So 
> it strikes me it would be quicker and less bug prone to just 
> re-implement the program in a modern language and gain a few 
> advantages along the way readability included.

There are many mainframes out there that don't run "modern" (i.e., PC-based) languages. Since Edgar is a mainframe consultant - at least so I gather from his bio - I can see why he'd be constrained to COBOL and such.

> It would have been nice if the original code/cobol code had been
> linked to or a summary or explaination of the programs function was 
> included.....

Well, I seriously doubt that Edgar's free to release his customer's code for everybody's perusal. I agree, it would be nice if the world ran on Open Source principles by default, and that it would indeed be interesting to see what the original programming task was - just so people can snark at the complexity and re-do it all in three lines of Perl or whatever - but that's not how things are. Until we take over, at least. :)

[[Jimmy]] - Ah, Cobol. Not the reason I dropped out of college (that was RPG), but a damn close second place.

[[Edgar]] - I won't try to one-up Ben's responses to the substantive issues. We readers know why we keep him around...

[[[Ben]]] - Thanks, Edgar - I appreciate the implied compliment.

[[Edgar]] - But essentially you and I have no dispute. Your objections/questions are all very reasonable. As I remarked to Ben, given a choice, I would prefer C over any other programming language I have ever used. Regarding the article, in retrospect it might have made sense to do things a bit differently. Some background.

Although I have been using Linux since SuSE 5.0, I still consider myself somewhat a newbie. Several decades of mainframe experience have inured me against the daily slings and arrows. But even a casual reading of my articles likely elicits no more than a polite "ho, hum" from the members of TAG. A while back I discovered how easy it is for even a newbie to do things that might sound intimidating, to a newbie. I'm addressing newbies.

[[[Ben]]] - Just to expand on that point a bit - that's one of the types of articles I'm always looking for. In fact, one of my larger challenges in vetting, editing, and writing articles is to keep or access that "beginner's mind" - and it's not an easy one. I'm always appreciative of new Linux users who can write well about their experiences - this is one of the most important types of articles in LG.

[[Edgar]] - As I was writing the article, I was concerned the problem itself might be getting too much "air time" as it were. But I doubt most newbies have done anything in Assembler. Any Assembler. And COBOL?

When I submitted the article, Ben properly chastised me for mentioning that 5-letter-word.

[[[Ben]]] - Ah-ah-ah! I did not. I did, however, poke some gentle fun at you about it. :)

[[Edgar]] - But I felt it necessary to motivate what I went through and to explain it in enough detail to make the problem understandable.

The point I really wanted to make -- and it may not have come across well -- was that Open Source might be of help where one wouldn't necessarily expect it and that it needn't be terribly difficult to install a package not part of the distribution you are using.

[[[Ben]]] - This is, in fact, how I took it; that's been a wonderful part of my Linux experience as well. The Open Source nature of it seems to spawn effort and creativity in people who wouldn't normally give a rat's ass about writing software, and motivates others who would otherwise let their ideas die without expression. As an example, some years ago I ran across a Linux tool that translated Norton Guide databases - something that made me very, very happy, since it allowed me to convert some very useful data I'd thought of as lost. Another author had rewritten the first IDE for C/C++ that I'd ever used under DOS (Borland's Turbo C++) as a Linux prog (RHIDE), which was responsible for a wonderful bout with nostalgia. :) There are lots of examples, many people doing wonderful things simply because the environment is available, there's nearly infinite room to play in, and you get full credit for your work.

[[Edgar]] - To that extent it might as well have been Pascal to C. Except that that wouldn't even deserve "ho, hum".

For precisely the reasons mentioned by Ben the code in the article is not even the code dealt with, although functionally equivalent. And long-term it is not a question of one, single program. This is far more than merely proof of concept. The people I am doing this with long ago left the starting blocks. And there is reason to believe that many other companies need quickly to get away from Assembler into something else main-stream. Anything! Before their last remaining Assembler programmer retires.

But, Trevor, you have given me a very interesting idea.

I need to check this out, but I am fairly confident that C is an intermediary step in the compilation process, under Linux. I'm the only one in this crowd regularly using Linux, but not C. The code might be ugly. I need to look into it. But down the road...?

Thanks for your comments.

[[[Ben]]] - I'll be looking forward to that article, too. :)

As to package installation - heck, most interesting things are already available as packages, especially in distros that make a point of it. When I find something on the Net that I'd like to install, I usually check for an existing Debian package containing it - and find it in better than half the cases.

ben at Fenrir:~$ apt-cache show open-cobol
Package: open-cobol
Priority: optional
Section: devel
Installed-Size: 464
Maintainer: Bart Martens <bart.martens at advalvas.be>
Architecture: i386
Version: 0.32-1
Depends: libc6 (>= 2.3.5-1), libdb4.3 (>= 4.3.28-1), libgmp3c2, libltdl3 (>= 1.5.2-2), 
	libncurses5 (>= 5.4-5), libcob1 (= 0.32-1), libcob1-dev (= 0.32-1)
Filename: pool/main/o/open-cobol/open-cobol_0.32-1_i386.deb
Size: 175364
MD5sum: e939fd76f9592030eabd051e8f168ba4
Description: COBOL compiler
 OpenCOBOL implements substantial part of the COBOL 85 and COBOL 2002
 standards, as well as many extensions of the existent compilers.  OpenCOBOL
 translates COBOL into C and compiles the translated code using GCC.
  Homepage: http://www.opencobol.org/

So for me, it's just a matter of typing "sudo apt-get install open-cobol". Debian takes care of all the dependencies, etc.

[[[[Trevor]]]] - Well guys I'd better explain what went through my mind when I read the article, it went something like "This guy's creating a problem' now my experience and current nightmare is legacy code, where I'm at (I work for IBM on uk government outsourcing projects) we have a number of older systems that need modified, updated and in many cases replaced. The cost of hiring people with 'legacy' skill sets which these days increasingly includes COBOL is high.

One day that piece of COBOL will need either it's function modified or the hardware will be replaced (nothing lasts for ever) with an off the shelf x86/opteron commodity server (hopefully running Linux) and someone will have to re-do the work and that might meaning having to hire an expensive contractor or re-write the whole system in some other language and that is a cost that could, at least in part, have been avoided.

I have seen huge amounts of time and money spent dealing with this type of 'patching up' it's my pet hate.

[[[[[Ben]]]]] - Sure; anyone who's been a programming consultant for any length of time has seen this, and it's always the same kind of double-bind. However, this is also exactly the situation I had in mind when I quoted the Golden Rule - what most of us neglect to consider, at least when we first run into the situatuon, is the financial/operations end of the problem from the client's point of view.

When a client (whose entire codebase consists of, say, COBOL) discovers that he needs to patch/modify/fix/tweak/whatever that codebase, the first questions that arise will inevitably be these:

1) What will it cost?

2) How much of an interruption of business will it create?

In the case of the first question, a patch - or even a large series of patches or mods - is usually much cheaper than a complete rewrite, especially when that codebase is large. So, the decision to patch, and keep patching some obscure horror that can only be fixed by one doddering wizard (all others having either croaked or switched to Java, which is essentially the same thing :) is pretty much the default - absent some immediately-pending and obvious (or already extant) catastrophe. Trying to convince clients that it's going to happen before that point will mark you as a doom-sayer with a sly second agenda, and will not result in happy relations with that client.

In the case of the second question, the problem is even worse - far worse. A patch - even if it's relatively major - can usually be rolled back fairly easily in case of problems; a complete replacement - which usually involves system changes (hardware, configuration, OS change) cannot. Certainly not easily. Can you, or any programmer, guarantee that bugs - even major ones - will not show up after the replacement, and take the business out of operation for some length of time? The answer is, of course, no - and so, a complete rewrite is very risky.

In short, clients usually hate rewrites, and love patches - and if you keep wishing that it wasn't otherwise, you'll only get frustrated. I learned, a long time ago, to take my satisfaction in charging all the market will bear for overtime, etc. when the crash does come; verbal "I told you so"s bring no more than transient satisfaction, and put no money in your pocket. :) By that time, I tend to be uniquely qualified as The Right Programmer to do the job - since I've used the patching process to study the client's entire business flow (as it relates to the software) and can not only write the new code but know how and where the system can be improved.

Looking at this kind of thing as an opportunity rather than a problem has, over time, proven to be very rewarding. :)

[[[[[Edgar]]]]] - Hi Trevor, I hope you don't feel that "we" are picking on you. This wasn't orchestrated. I still have essentially your attitude but a few years ago resigned myself to the fact that the real world is as described by Ben. About 5 years ago I quit telling everyone who might understand that we need to get away from Assembler. As an outside consultant I don't have access to the appropriate managers and they wouldn't understand the problem anyhow.

Basically I'm as frustrated as you are. But I have learned to live with it and don't lose any sleep over the situation. That the managers are incompetent is clear when you consider that I am now the only one left with Assembler skills. An outsider! Unexpectedly, I spent a couple of weeks off work this year due to skin cancer (taken care of) and during that time absolutely nothing happened at work, in spite of the fact that we had a deadline of 31 March. How can they sleep at night if they understand the situation?!

That is my major customer at the moment, government service. I can't speak about the UK but I am now convinced that the government service systems in Germany and the USA cannot be reformed and need to be junked. I'm not going to start a campaign and it of course won't happen. But if the tax-payers were aware of the money wasted and the virtual sabotage on the part of too-highly-rated civil servants...

In the case of converson to COBOL, that is a project for someone I have twice worked for and have known for well over 30 years. The customer, for whatever reasons, wants to move from Assembler to COBOL. My guess is that the rest of what they have is in COBOL and if that in fact is the case, the decision is not completely wrong. At least it is a simplification of their computing environment.

By the way, in both cases I am dealing with financial computing, i.e. BIG mainframes. Until discovering Open COBOL and GMP I wasn't aware that decimal arithmetic was readily available outside the mainframe world. Well, back in 8080 and Z-80 days I used DAA, 'nuff said.

I'm afraid, as much as I don't want to admit it, Ben once again is right. The world just ain't the way we'd like it. When I did my military service there was an expression: shit floats. Basically Peter was an optimist, if only incompetence were readily recognizable at the level where it first floats! But somehow, perhaps as it rots and begins to stink, it seems to take up more space and float even higher.

Sadly, too many people, like Billy, could sell SNOW to Eskimos.

Don't give up trying to do things right. But don't let it get to you psychologically.


Talkback: Discuss this article with The Answer Gang

Published in Issue 128 of Linux Gazette, July 2006


By Howard Dyckoff



Please submit your News Bytes items in plain text; other formats may be rejected without reading. [You have been warned!] A one- or two-paragraph summary plus a URL has a much higher chance of being published than an entire press release. Submit items to bytes@linuxgazette.net.

News in General

boltEclipse Callisto shines in FOSSw world

"Callisto" is a coordinated, simultaneous release of some 10 new or upgraded releases of major projects. This includes version 3.2 of the Eclipse Platform and these other new releases:

* Business Intelligence and Reporting Tools (BIRT) Project
* C/C++ IDE
* Data Tools Platform
* GEF - Graphical Editor Framework
* Eclipse Project [v. 3.2]
* Eclipse Test and Performance Tools Platform Project
* Eclipse Web Tools Platform Project
* VE - Visual Editor and
* Eclipse Modeling and Graphical Frameworks

Callisto was a major undertaking for the Eclipse community, involving 10 different project teams, 260 committers and over 7 million lines of code. Demonstrating the multi-vendor and global nature of the Eclipse community, 15 different ISVs contributed open source developers to work on the projects included in Callisto.

The coordination took the better part of a year and involved Bjorn Freeman-Benson, technical director of infrastructure at the Eclipse Foundation, and Ward Cunningham, Director, Committer Community Development, who left Microsoft in 2005 to work with Eclipse. [Cunningham has been described as the father of the wiki.] Eclipse presenters from IBM at JavaOne had described the level of maturity and coordination built into the Eclipse process during a morning keynote and also at a JavaOne BOF. This is based in part on advanced collaboration tools shared by all Eclipse project leads [IBM suggested that it may productize some of the process tools]. It also involved coordinating bug fixes with Bugzilla.



Also see free archived webinars on how you can use the different Eclipse projects. These use Adobe's Macromedia Breeze meeting service to host the webinars: http://www.eclipse.org/callisto/webinars.php

boltRSA folds into EMC

EMC had an expensive lunch at the end of June. For a whopping $2.1 billion, it acquired security and identity company RSA, orignially founded by the 3 cryptologists whose initials make up its name.

Although financial analysts were skeptical about the potential integration issues and the high cost, EMC's CEO Joe Tucci called the purchase "critical technology" for adding security to storage solutions. One area of possible synergy is regulatory compliance, where data security is of increasing importance. However, RSA has not been primarily a sales organization, as EMC is, and the corporate cultures may clash.

RSA put itself up for auction several months ago, according to NY Times reports.

bolt Red Hat Enterprise Linux 5 beta starting in July

Beta 1 of the next RHEL server may to be available in late July. The now-available Fedora Core 5 is viewed as alpha code for RHEL5. Red Hat hopes to incorporate the next Fedora Core 6 build into its Beta 2, planned for release in mid September.

RHEL 5 will include the Xen open source virtualization hypervisor -- a key element of the release and a necessary addition due to Novell's inclusion of Xen in Suse 10 -- as well as integration with the Red Hat's directory and certificate servers. It will also offer support for Intel and AMD virtualization extensions, stateless Linux clients, and single sign-on.

boltJBoss Unveils OSSw Enterprise Management Strategy with implementation of agent technology

JBoss, now a division of Red Hat, is open sourcing the core systems management agent in JBoss Operations Network (ON) to create and drive broadened adoption and collaboration around its open management platform. This announcement, backed by the JBoss developer and ISV community, provides software vendors and customers a foundation for building their own management agents, which will enable interoperability across heterogeneous IT environments.

This strategy is an important step in putting customers in charge of their IT infrastructures. As a subscription-based offering, JBoss ON delivers dramatic savings in total cost of ownership (TCO) and provides a holistic environment for inventory management, administration and configuration, monitoring, software updates and provisioning of applications based on JBoss Enterprise Middleware Suite (JEMS). Red Hat is working with other network and systems management vendors to ensure that management data can be shared with existing customer management installations. JBoss ON subscribers will also be able to leverage existing open source agents from projects such as Nagios.

By bringing its JBoss management agent to the open source community, Red Hat will enable systems to expose its manageability functions in an extendable and pluggable way - underscoring the company's commitment to support heterogeneous customer environments. For example, other vendors can extend the agent to manage their products. As part of its strategy, JBoss will create blueprints, certification toolkits and methodologies that vendors can use to validate their extensions and plug into the management process used for JEMS today.

boltJBoss Releases JBoss Seam 1.0, a Web 2.0 framework for SOA technologies

JBoss also announced the general availability of JBoss Seam 1.0, a new application framework for Web 2.0 applications that unifies popular service-oriented architecture (SOA) technologies such as Asynchronous JavaScript and XML (AJAX), JavaServer Faces (JSF), Enterprise JavaBeans (EJB) 3.0, Java portlets, business process management (BPM) and workflow. Since its initial developer release, JBoss Seam has seen strong community interest and has played a driving role for a new standards initiative for Web Beans through the Java Community Process (JCP).

Designed to eliminate complexity at the architecture and application programming interface (API) level, JBoss Seam enables developers to assemble complex web applications with simple annotated POJOs (plain old Java objects), componentized UI widgets and simple XML. To accomplish this, JBoss Seam extends the annotation-driven and configuration-by-exception programming model of EJB 3.0 into the entire web application stack. It bridges the artificial gap between EJB 3.0 and JSF in the Java Platform Enterprise Edition 5.0 (Java EE 5.0) architecture. The result is a unifying, tightly integrated application model that enables stateless, stateful, transactional and process-driven applications such as workflow and page flow.

Gavin King, founder and project lead of JBoss Seam and founder of Hibernate, commented: "Enabling the next generation of web development requires a major reconsideration of the underlying web application architecture. Until EJB 3.0, that had not been possible. As the first unifying ... framework for SOA technologies, JBoss Seam offers developers a rapid development environment and programming model that extends from the simple to the most complex web applications."

Key features of JBoss Seam 1.0 include:

-- EJB-based development. EJB 3.0 has changed the notion of EJBs as coarse-grained, heavy-weight objects to EJBs as lightweight POJOs with fine-grained annotations. In JBoss Seam, everything is an EJB. JBoss Seam embraces the Web 2.0 concept that the web is the platform, and as such, JBoss Seam eliminates the distinction between presentation tier components and business logic components. Even session beans, for example, can be used as JSF action listeners.
-- AJAX-based remoting layer. JBoss Seam Remoting allows EJB 3.0 session beans to be called directly from the web browser client via AJAX. The session beans appear as simple JavaScript objects to the JavaScript developer, hiding the complexity of XML-based serialization and the XMLHttpRequest API. Web clients may even subscribe to JMS topics and receive messages published to the topic as asynchronous notifications.
-- Declarative state management for application state. Currently, Java EE applications implement state management manually, an approach that results in bugs and memory leaks when applications fail to clean up session attributes. JBoss Seam eliminates almost entirely this class of bugs. Declarative state management is possible because of the rich context model defined by JBoss Seam.
-- Support for new types of stateful applications. Before Seam, HTTP session was the only way to manage web application states. JBoss Seam provides multiple stateful contexts of different granularity. For example, developers can write web applications with multiple workspaces that behave like a multi-window rich client.
-- Support for process-driven applications. JBoss Seam integrates transparent business process management via JBoss jBPM, making it easier than ever to implement complex workflow and page flow applications. Future versions of JBoss Seam will allow for the definition of presentation-tier conversation flows by the same means.
-- Portal integration. JBoss Seam supports JSR-168 compliant portals such as JBoss Portal.

JBoss Seam 1.0 is free to download and use under the GNU Lesser General Public License (LGPL). JBoss Seam 1.0 works with any application server that supports EJB 3.0, including JBoss. For download and additional information, visit www.jboss.com/products/seam

boltWebinar Series on Clustering JVM, Tomcat Sessions

Terracotta, Inc., announced a free Webinar series for Java professionals about the technology and benefits of the company's clustered JVM (Java Virtual Machine) solutions The series is a response to the market's growing interest in clustering at the JVM level, instead of at the application level.

Designed for Java developers, architects, and anyone needing fault tolerance and linear scalability, the webinars will cover all aspects of clustering the JVM. Highlights will include benefits, new features, installation, usage, management, demos, and technical Q&As. Webinars will be archived for those who cannot attend.

The first webinar, on June 28, addressed clustering the JVM with Apache Tomcat, which Terracotta currently supports. The Panelists included Jim Jagielski, CTO, Covalent Technologies; Gary Nakamura, vice president, Terracotta; and Sreeni Iyer, senior manager/field engineer, Terracotta

Terracotta Sessions clusters the JVM, instead of the application, so no coding or tuning during development is needed while achieving high availability and linear scalability. Developers can easily replicate sessions to other servers for high availability, without sacrificing performance. The product is fee for developers.

Future Webinars will focus on the Spring Framework from Interface21, which Terracotta will support in Q3, and additional open source projects that Terracotta will support. Register at : http://www.terracottatech.com/

boltImproving Make with Mr. Make

If you want to expand your software build repertoire, check out a 4-part series of practical techniques for enhancing your GNU Make Makefiles. John Graham-Cumming presents "recipes" for improving your Make-based builds. For example, Recipe 4 shows how to determine the version of GNU Make in a Makefile, or if a specific GNU Make feature is available. Recipe 5 performs a recursive $(wildcard), and Recipe 6 demonstrates tricks for tracing rule execution.

Interested? See: http://cmc.unisfair.com/log_thru.jsp?eid=72 and http://cmc.unisfair.com/index.jsp?id= 755&code= mrmake [for part 1, use id=844 for part 2 and id=993 for part 3]. The last part will occur in August.

boltOpera 9 is out [and eWeek likes it]



New Opera widgets make for small, light-weight applications that execute locally in the browser. These include calendar and game widgets, currency converters, etc.

Opera's expanded functionality makes it possible to selectively block content within a Web site. This can be done with right-clicking on the site page, then holding down the Shift key and clicking on specific images, ads and other componentsto be blocked on the next visit.

Quoting from the eWeek review: "A new site-settings feature made it possible to define controls and settings on a site-by-site basis. So, for example, we could define how we wanted to deal with pop-ups or cookies on a site.

"Now integrated directly within the Opera browser is a BitTorrent client, useful for downloading very large files (legal and non-legal). This client worked very well in our tests, and during downloads of Linux ISO files it provided good feedback and was very lightweight. However, BitTorrent users should keep in mind that if you close the browser, the download stops. Many BitTorrent clients, in contrast, just switch to a minimized mode. "

Here's a full list of new features: http://www.opera.com/products/desktop/

Conferences and Events

==> All LinuxWorld Expos < ==
Dr. Dobb's Architecture & Design World 2006
July 17-20, Hyatt Regency McCormick Place Conference Center, Chicago, IL
O'Reilly Open Source Convention 2006
July 24-28, Portland, OR
August 30-- 01 September, Birmingham, U.K
7/30 - 8/03, Moscone Center, Boston, MA
Entertainment Media Expo 2006
8/07 - 8/09, Universal City, CA
LinuxWorld Conference & Expo -- SF
August 14-17, 2006 -- in foggy San Francisco, dress warmly!!
SD Best Practices 2006
September 11-14, Hynes Convention Center, Boston, MA
GridWorld 2006
September 11-14, Convention Center, Washington, DC.
Digital ID World Conference
September 11-13, 2006, Santa Clara Marriott, Santa Clara, CA.

[The Dig-ID Conference sessions are on areas such as: enterprise identity management, provisioning, strong authentication, federated identity, virtual directories, smart cards, web services security, identity-based network access control, enterprise rights management, and trusted computing. I found the 2005 conference to be most excellent. Further information on Digital ID World 2006, is here: http://conference.digitalidworld.com/2006/]

Distro news

The latest stable version of the Linux kernel is: [ http://www.kernel.org/pub/linux/kernel/v2.6/patch- ]

boltXandros Desktop 4 released

Xandros Corporation has announced immediate availability of Xandros Desktop 4, a novice-friendly desktop Linux distribution based on Debian GNU/Linux: "Xandros, the leading provider of easy-to-use Linux alternatives to Windows desktop and server products, today announced a new line of consumer desktop products targeting home and multimedia users: Xandros Desktop Home Edition and Xandros Desktop Home Edition - Premium."

Xandros Desktop Home Editions provide enhanced Windows compatibility, including enhanced Windows-to-Linux migration support. This is the first distribution to enable writing to NTFS partitions (the native file system on Windows computers) from Linux, which allows users to work with the same files from Linux and Windows.

boltAnnouncing Fedora Core 6 Test 1 (5.90)

The Fedora Project announced the first release of the Fedora Core 6 development cycle, available for the i386, x86_64, and ppc/ppc64 architectures, including Intel based Macintosh computers. Beware that Test releases are recommended only for Linux experts/enthusiasts or for the technology evaluation, as many parts are likely to be broken adn the rate of change is rapid. Test 2 is scheduled for release July 19, marking the developmental freeze of the Fedora Core 6 release. No new features after this point. It is important that we get your help in testing, reporting and suggesting fixes for bugs, and directing the technological improvements we attempt with this release of Fedora Core.

See info at : http://fedoraproject.org/wiki/Core/Schedule

boltSimplyMEPIS 6.0 Release Candidate 2 adds Security

The MEPIS team released the second release candidate of SimplyMEPIS 6.0, on June 21. RC2 adds bug fixes, security updates, and screen resolution detection, founder and lead developer Warren Woodford said. The distribution now also includes monitor resolution autodetection.

[Monitor resolution autodetection] is a feature that has been requested for a long time," according to Woodford. "If the user does not specify their desired display resolution when booting the CD, we attempt to obtain it from the EDID data returned by the display card, and then the optimum screen resolution is chosen automatically. Unfortunately there are a lot of monitor-display card combinations that do not return this data, so this feature will work only for some users."

A bug was found in RC1 that broke the "install on entire disk" option, Woodford said, and it has been fixed for RC2. By popular demand, the apt-notify applet was improved to support transparency in the panel. Firefox now runs much faster out-of-the-box, due to some configuration improvements suggested by the community, he added.

RC2 incorporates more security updates from the Ubuntu Dapper pool, including a security-patched version of the 2.6.15 kernel

boltSUSE Linux Enterprise Release Candidate 3

Novell has announced the availability of SUSE Linux Enterprise 10 RC3, both Desktop (SLED) and Server (SLES) editions, for public testing: "Be among the first to install, test and enjoy SUSE Linux Enterprise 10. The pre-release contains all the functionality of the regular release, but is not the final product. SUSE Linux Enterprise 10 is an open, flexible and secure platform that is ready to host the applications and databases critical to your business -- from the desktop to the data center, across a wide variety of workloads."

See: http://www.novell.com/linux/preview.html

boltOpenOffice Security Bulletin [06-29]

OpenOffice.org 2.0.3 fixes three security vulnerabilites that have been found through internal security audits. Although there are currently no known exploits, we urge all users of 2.0.x prior to 2.0.2 to upgrade to the new version or install their vendor's patches accordingly. Patches for users of OpenOffice.org 1.1.5 will be available shortly. [note: the Macro execution flaw allows an attacker to get a macro executed even if the user has disabled document macros.]

The three vulnerabilities involve:
* Java Applets, CVE-2006-2199
* Macro, CVE-2006-2198; and
* File Format, CVE-2006-3117

Security News

"Blue Pill = Security Nightmare"

From Joanna Rutkowska on the 'theinvisiblethings' blog:

"Over the past few months I have been working on a technology code-named Blue Pill, which is just about that - creating 100% undetectable malware, which is not based on an obscure concept.

"The idea behind Blue Pill is simple: your operating system swallows the Blue Pill and it awakes inside the Matrix controlled by the ultra thin Blue Pill hypervisor. This all happens on-the-fly (i.e. without restarting the system) and there is no performance penalty and all the devices, like graphics card, are fully accessible to the operating system, which is now executing inside virtual machine. This is all possible thanks to the latest virtualization technology from AMD called SVM/Pacifica."

See the full entry at http://theinvisiblethings.blogspot.com/2006/06/introducing-blue-pill.html

boltSelected Security NEWS from SANS Institute

--Buffer Overflow Flaw in Opera Browser (23 June 2006)

A buffer overflow flaw that occurs when the Opera web browser processes JPEG mages could allow remote code execution. The problem is known to exist in Opera v.8.54 and possibly in earlier versions as well. Users are urged to upgrade to the new Opera v.9. http://www.vnunet.com/vnunet/news/2158971/jpeg-flaw-uncovered-opera

--DATA THEFT & LOSS --Lost Memory Stick Holds Phishing Investigation Dossier (26 June 2006)

A police officer with the Australian High Tech Crime Centre (AHTCC) lost a memory stick that contains sensitive financial data belonging to thousands of Australians. The lost memory stick holds a dossier on Russian phishing scams. The data on the stick were being used in an investigation; several arrests were made with the help of the data, but since the loss of the stick, no arrests have been made. While officials searched fruitlessly for the memory stick, the people whose data were compromised were not informed of the loss. The officer who lost the device violated AHTCC rules regarding data transport. http://australianit.news.com.au/common/print/0,7208,19588463%5E15306%5E%5Enbv%5E,00.html

--Attackers Use SMS Messages to Lure People to Malicious Site (23 June 2006)

A recently detected attack sends intended victims SMS text messages thanking them for subscribing to an online dating service and telling them they will be charged US$2 a day until they unsubscribe. When people visit the site where they are purportedly unsubscribing from the fictitious service, "they are prompted to download a Trojan horse program." Infected computers then become part of a botnet. http://www.zdnet.co.uk/print/?TYPE=story&AT=39277240-39020375t-10000025c

--Survey Finds Americans Want Strong Data Security Legislation

A survey from the Cyber Security Industry Alliance (CSIA) of 1,150 US adults found 71 percent want the federal government to enact legislation to protect personal data similar to California's data security law. Of that 71 percent, 46 percent said they would consider a political candidate's position on data security legislation and "have serious or very serious doubts about political candidates who do not support quick action to improve existing laws." In addition, half of those surveyedavoid making online purchases due to security concerns.


--Millions of Blogs Inaccessible Due to DDoS Attack

A "massive" distributed denial-of-service (DDoS) attack on Six Apart's blogging services and corporate web site left about 10 million LiveJournal and TypePad blogs unreachable for hours on Tuesday, May 2.

Six Apart plans to report the attack to authorities.


--Soon-to-be-Proposed Digital Copyright Legislation Would Tighten Restrictions

Despite efforts of computer programmers, tech companies and academics to get Congress to loosen restrictions imposed by the Digital Millennium Copyright Act (DMCA), an even more stringent copyright law is expected to be introduced soon. The Intellectual Property Protection Act of 2006 would make simply trying to commit copyright infringement a federal crime punishable by up to 10 years in prison. The bill also proposes changes to the DMCA that would prohibit people from "making, importing, exporting, obtaining control of or possessing" software or hardware that can be used to circumvent copyright protection.


Software and Product News

boltNew Berkeley DB Java Edition 3.0

The new Berkeley DB 3.0 is a high performance, transactional storage engine written entirely in Java. Like the original, highly successful Berkeley DB product, Berkeley DB Java Edition executes in the address space of the application, without the overhead of client/server communication. It stores data in the application's native format, so no runtime data translation is required. Berkeley DB Java Edition supports full ACID transactions and recovery. It provides an easy-to-use, programmatic interface, allowing developers to store and retrieve information quickly, simply and reliably.

Berkeley DB Java Edition is designed to offer the same benefits of Enterprise Java Beans 3.0 (EJB3) persistence without the need to translate objects into tables.

Most persisted object data is never analyzed using ad-hoc SQL queries; it is usually simply retrieved and reconstituted as Java objects. The overhead of using a sophisticated analytical storage engine is wasted on this basic task of object retrieval. The full analytical power of the relational model is not required to efficiently persist Java objects. In many cases, it is unnecessary overhead. In contrast, Berkeley DB does not have the overhead of an ad-hoc query language like SQL, and so does not incur this penalty.

The result is faster storage, lower CPU and memory requirements, and a more efficient development process. Despite the lack of an ad-hoc query language, Berkeley DB Java Edition can access Java objects in an ad-hoc manner, and it does provide transactional data storage and indexed retrieval, as you would expect from any database. The difference is that it does this in a small, efficient, and easy-to-manage package. Using the Persistence API, Java developers can quickly and easily persist and retrieve inter-related groups of Java objects.

Berkeley DB Java Edition was designed from the ground up in Java. It takes full advantage of the Java environment. The API provides a Java Collections-style interface, as well as a programmatic interface similar to the Berkeley DB API. Its architecture supports high performance and concurrency for both read- and write-intensive workloads.

Berkeley DB Java Edition is not a relational engine built in Java [like Derby]. It is a Berkeley DB-style embedded store, with an interface designed for programmers, not DBAs. The architecture is based on a log-based, no-overwrite storage system, enabling high concurrency and speed while providing ACID transactions and record-level locking. Berkeley DB Java Edition efficiently caches most commonly used data in memory, without exceeding application-specified limits.

Here's the download link : http://dev.sleepycat.com/downloads/releasehistorybdbje.html

boltAjax4jsf Now open source

Exadel, Inc. has contributed Ajax4jsf as an open source project to Java.Net, the web-based, open community that facilitates Java technology collaboration in applied areas of technology and industry solutions.

Exclusively created to bring rich user interface functionality to the JavaServer Faces (JSF) world, Ajax4jsf is a rich component development framework that extends the benefits of JSF to AJAX development. Ajax4jsf allows developers to add AJAX-capability to existing JSF applications, create rich components with built-in AJAX support, create a slick user interface with Skinning technology, package Java Script files and other resources together with JSF components, and test the components functional elements in the process of components development.

More information about Ajax4jsf can be found at: http://ajax4jsf.dev.java.net

boltUnisys Announces Open Source 'Oasis' in Enterprise Computing

Unisys Corporation has announced a major expansion of its open source capabilities with Unisys Open and Secure Integrated Solutions (OASIS), an integrated, certified set of software suites targeted specifically for enterprise class computing based on open standards and fully supported through Unisys global services and vertical-industry solutions capabilities. Combining technology and services, Unisys open source solutions provide a comprehensive environment that realizes the full benefits of enterprise computing capitalizing on the economics of OSSw.

Unisys OASIS and associated services enable clients to take a disciplined, secure approach to realizing the cost, agility and security advantages that open source solutions can provide. The Unisys OASIS software suites provide a range of options, including:

-- Application Server Suites, to migrate proprietary Java Enterprise Edition environments to open source. Suites include JBoss Application Server, a custom-tuned Java Virtual Machine for use with Unisys scaleable server platforms; and Unisys Application Defender, which increases the capability of Java Web applications in the JBoss environment to withstand attacks by hackers and other adversaries.
-- Database Server Suites, which integrate and deploy the open source databases MySQL and PostgreSQL.
-- Solutions for rapid migration to an open source platform, which include tooling, services and methodology for moving databases and application servers.
-- Services for developing Unisys 3D Visible Enterprise (3D-VE) blueprints, which help organizations map their business to the IT infrastructure, enabling evolution toward a Service Oriented Architecture (SOA).

For additional information on Unisys OASIS, visit http://www.unisys.com/services/open__source.htm .

boltBlueCat Networks and Mirage Networks Defend Network Access

BlueCat Networks and Mirage Networks announced that the companies have formed an alliance to deliver the strongest DHCP-based network access control (NAC) available. Through this partnership, BlueCat Networks is adding support for Mirage NAC to its lineup of network appliances. The combined solution will enable customers to cost-effectively authenticate users and control their access to network resources.

A recent Gartner report, states "To protect their networks, many network managers are implementing network access control (NAC), based on technologies such as 802.1x and Dynamic Host Configuration Protocol (DHCP)." This reflection of industry requirements and drivers has spurred the development of this partnership.

BlueCat selected Mirage NAC for its abilities to provide a future-proof approach that provides security no matter the IP device or operating system. Its award-winning design offers complete access control by:

-- Addressing policy violations and threats -- even day zero -- at all points on the network
-- Enforcing DHCP-based authentication for dynamic IP addresses
-- Providing safe quarantining and mitigation of offending devices

"There are no safe havens in the network: every endpoint, from the network printer to the VoIP phone to the Windows desktop is vulnerable to threats," stated Michael D'Eath, vice president of corporate and business development, Mirage Networks. "BlueCat is clearly a leader in the IPAM, DNS and DHCP Space. With this partnership, BlueCat can securely authenticate and attach any endpoint, managed or unmanaged, to the network, without the need for expensive switch upgrades, cumbersome client software or reactive signature updates."

The joint solution is available through BlueCat Networks. Contact BlueCat Networks at www.bluecatnetworks.com or 1-866-895-6931 for pricing and more information.

boltForum Systems Launches Web Services Security Software Development Kit

Forum Systems has released the Forum Java Web Services Security (JWSS) Software Development Kit (SDK) version 1.0 offering developers a comprehensive library of application programming interfaces (API's) to leverage in coding Web Services applications. The JWSS SDK v1.0 addresses the need for security to be enforced within the application itself in order to ensure privacy and integrity of Web Services and SOA applications.

Forum JWSS SDK v1.0 allows developers to administer and apply security policies within J2EE compliant Application Servers using a declarative security model and API's for XML message authentication and authorization. The Application Server is then responsible for applying these security constraints to the code at runtime. Adding XML security within business logic prevents transactions from bypassing third-party enforcement points that would violate regulatory compliance or work flow security. Companies should complement this developer-centric approach with an "interception" model using a SOA Gateway or XML Firewall within a DMZ (Demilitarized Zone) for a global and scalable SOA security and acceleration strategy.

"There are business applications in which security must be enforced at the exact location where information is being processed for privacy reasons," said Walid Negm, vice president of marketing for Forum Systems. "In certain financial services and government settings, the origin of messages must be verified at the point of consumption to ensure that they were not physically intercepted. When using XML Encryption within the application, Forum's JWSS SDK v1.0 equips the developer with XML security functionality to decipher and verify the message contents," Negm added.

Forum Seamless Security Solutions Architecture (Forum S3A) is an adaptive approach to building security-minded service-oriented applications and data-level networks using life-cycle solutions including vulnerability management, testing systems, firewalls and gateways. Forum products are available as software, PCI-card and appliance options and comply with government requirements including CheckPoint OPSEC Certification, FIPS Certification, Common Criteria EAL 4+ (in process) and JITC DoD PKI Certification. Forum Systems is an active a member of OASIS and WS-I helping mature standards such as WS-I Basic Profiles, SAML and WS-Security.

Magical Realism... (non-Linux news of general interest)

bolt WOMMA Site Devoted to Word of Mouth Research and Measurement

The Word of Mouth Marketing Association (WOMMA) has launched the first site devoted solely to the field of word of mouth (WOM) research. Featuring a new blog and email newsletter, the WOM Research site is a resource for both marketers and academics seeking the latest information on word of mouth measurement and metrics. Everything can be found at http://www.womma.org/research.

The new site features content designed to introduce WOM research to a wider audience. In addition to the latest research reports, surveys, and data, the WOM Research blog also features original contributions from leading academics and market research experts.

WOMMA has also announced the Word of Mouth Basic Training 2 Conference to be held on June 20-21 in San Francisco. Measurement and research will form a significant part of the agenda, with research experts on hand from BIGresearch, Biz360, Informative, Keller Fay Group, Millward Brown, MotiveQuest, Starcom, VoodooVox, and Umbria. More details can be found at http://www.womma.org/wombat2.

As word of mouth marketing has continued to explode into the mainstream marketing mix, the demand for new resources and data has become essential. According to eMarketer, 43% of all marketers will incorporate word of mouth into their marketing programs in 2006.

WOMMA is the official trade association for the word of mouth marketing industry. Its members are committed to building a prosperous word of mouth marketing profession based on best practices, measurable ROI, and ethical leadership. Learn about WOMMA at http://www.womma.org.

boltWorld's Record Etch A Sketch Coming to SIGGRAPH 2006

The world's largest Etch A Sketch(R) will make its debut at SIGGRAPH 2006, the International Conference and Exhibition on Computer Graphics held 30 July to 3 August 2006 in Boston, Massachusetts. SIGGRAPH 2006 will bring an estimated 25,000 computer graphics and interactive technology professionals from six continents the event.

Officially endorsed by The Ohio Art Company, the installation will be in use before the curtain rises on the SIGGRAPH Computer Animation Festival - the world's marquee showplace for the latest and most innovative animation films of the year.

It allows audience members to control (in real-time) the two famous Etch A Sketch(R) drawing knobs and use them interactively on the main projection screen. Functionality will also include the audience's ability to "shake" the screen clean and start again with a blank canvas.

The SIGGRAPH 2006 Computer Animation Festival features approximately 100 films and videos by some of the world's most creative scientists, animators, VFX specialists, educators, studios, and students. For more information on the Computer Animation Festival,visit: http://www.siggraph.org/s2006/main.php?f=conference&p=caf.

boltVirtualization Software Allows SPARC Applications to Run on Linux

Transitive Corp recently announced the first two products in its series of products that are being developed as part of its new Solaris/SPARC Migration Initiative:

-- QuickTransit for Solaris/SPARC-to-Linux/Xeon and
-- QuickTransit for Solaris/SPARC-to-Linux/Itanium

Transitive Corp previously developed Rosetta which allows Macintosh applications written for the PowerPC chip to run on the new Intel-based versions of the Macintosh.

The new products are the first results of the collaboration with Intel announced in March of this year to accelerate the migration from RISC-based platforms to Intel-based platforms, and they allow Solaris/SPARC applications to run without any source code or binary changes on Linux/Xeon- and Linux/Itanium 2-based servers respectively. The products dramatically reduce the barriers IT organizations face when migrating Solaris/SPARC workloads to Linux/Xeon- and Linux/Itanium-based servers. Future products will allow SPARC applications to run on various other hardware platforms.

"One of the biggest obstacles to migrating from one server to another is getting your internally developed and commercial applications ported to the new server," said Bob Wiederhold, President and CEO of Transitive Corporation. "In a large IT organization this can take years to accomplish. Since SPARC applications can now run on your target platform without the burden of porting, this migration barrier is eliminated and IT organizations can immediately realize the full benefits of their new strategic server platform."

Transitive's QuickTransit hardware virtualization products allow SPARC applications to run with full functionality, including interactive and graphics performance. The use of QuickTransit is completely transparent to the end-user and easily managed by IT system administrators. Commercial and in-house software development teams can easily support their applications with QuickTransit because no source code or binary changes are required.

Transitive expects to deliver the first product, QuickTransit for Solaris/SPARC-to-Linux/Xeon, in Q3 2006. The second product, QuickTransit for Solaris/SPARC-to-Linux/Itanium, is expected to be available by the end of the year 2006.

boltAMD to gradually introduce DDR3, FB-DIMM; will be used in next gen Cray supercomputer

In a recent supercomputing win, AMD Opteron processors were selected for a multi-year contract that Cray, Inc. signed with Oak Ridge National Laboratory (ORNL) to provide the world's first petaflops-speed (1,000 trillion floating-point operations/second) supercomputer. The contract calls for progressive upgrades to ORNL's existing Cray XT3™ supercomputer, starting with Next-Generation AMD Opteron processors with DDR2 memory, followed by upgrades to use quad-core AMD Opteron processors, which will be socket compatible. These upgrades will accelerate peak speed to 250 teraflops (250 trillion floating-point operations per second), planned in late 2007.

ORNL is then expected to install a next-generation Cray supercomputer in late 2008. This system, currently code-named 'Baker,' is designed to deliver peak performance of one petaflops, making it roughly three times faster than any existing computer in the world. All systems provided for in the contract will utilize current and future versions of the AMD Opteron processor.

At the May Processor Forum, Senior Fellow Chuck Moore described a gradual process of innovation for AMD multi-core chips. This involved supporting advanced RAM technology as the price points come down near current DDR2 pricing.

At the presentation, AMD indicated that new architecture products from Intel will roughly halve the power consumption advantage AMD now enjoys. AMD will add better power management to keep its lead on future products in the 2007 and 2008 timeframe.

AMD also hopes to improve HyperTransport technology to handle up to 5.2 gigatransfers per second. AMD is encouraging motherboard manufacturers to add HTX slots that use the high speed data transfers.

AMD has plans to add an on-chip L3 cache, shared by all CPU cores, while maintaining the localized L2 caches for each core. Depending on its size and the use of pre-fetch algorithims, this feature would keep CPU cores running longer between pipleline stalls.

AMD will simplify its product portfolio by dropping the 939 and 754 processors in July and will offer steep discounts of up to 46% to stay ahead of Intel on chip pricing.

Meanwhile, Intel may have delayed the anticipated shipment of its 'Conroe' Duo and Extreme processors by up to four days, according to Taiwanese manufacturers. [ But the delay was not officially confirmed by Intel. ] The graphics chipsets to support the CPUs will also have a July 27th ship date [and this may be the cause]. Also, its entry-level Celeron D 360 CPU, now the last part to use the Pentium 4 NetBurst microarchitecture, will be introduced on September 1. [So there may be a back-to-school price war in September.]

boltWikipedia presents: Alien insults for rookies


The next time a Klingon shouts that your mother has a smooth forehead, you will know how to insult him back, thanks to a Wikipedia list of fictional expletives — curses and insults from books, TV series and movies, mainly science fiction and fantasy. [ Check it out at : http://en.wikipedia.org/wiki/List_of_fictional_expletives]

boltFree file backup and synchronization online with MediaMax

Streamload, a leading provider of online digital media services, today announced Streamload MediaMax 1.5, the latest version of its online media center. The new Streamload MediaMax service is the first in the online storage space offering a full suite of media sharing and remote access applications that include free file backup and file synchronization in one online service.

Streamload MediaMax allows users to easily store, organize, access, and share their digital media collection . Version 1.5 integrates the management and sharing of online and offline media by integrating free client software, called Streamload MediaMax XL, into the Streamload MediaMax service using a drag and drop window on their desktop. It also automatically performs regular backups of designated folders from their PC and it synchronizes files across multiple computers and devices every time new files are added, changed or deleted on their computer. A benefit of MediaMax's automatic backup and sync is always having files when and where they are needed without the hassle of manual uploads.

Users of the service can also invite friends and family to sync folders on their computers. Without any additional effort, users can automatically share photos and personal video instantaneously with others who are subscribed to their synchronized folders.

Streamload MediaMax runs on Windows, Mac OS and Linux. The standard service is free and comes with 25 GB of online storage. A premium account gives subscribers 250 GB of storage for $9.95 per month, and the elite subscription offers 1000 GB for $29.95 per month (when paid annually.) For more information and to download the free Streamloader desktop software visit http://www.mediamax.com.

boltScientists OK Gore's movie for accuracy / USA Today

[from -- http://www.usatoday.com/tech/science/2006-06-27-gore-science-truth_x.htm]

The nation's top climate scientists are giving "An Inconvenient Truth", Al Gore's documentary on global warming, five stars for accuracy.

The former vice president's movie — replete with the prospect of a flooded New York City, an inundated Florida, more and nastier hurricanes, worsening droughts, retreating glaciers and disappearing ice sheets — mostly got the science right, said all 19 climate scientists who had seen the movie or read the book and answered questions from The Associated Press.

The AP contacted more than 100 top climate researchers by e-mail and phone for their opinion. Among those contacted were vocal skeptics of climate change theory. Most scientists had not seen the movie, which is in limited release, or read the book.

But those who have seen it had the same general impression: Gore conveyed the science correctly; the world is getting hotter and it is a manmade catastrophe-in-the-making caused by the burning of fossil fuels.

Talkback: Discuss this article with The Answer Gang

Bio picture Howard Dyckoff is a long term IT professional with primary experience at Fortune 100 and 200 firms. Before his IT career, he worked for Aviation Week and Space Technology magazine and before that used to edit SkyCom, a newsletter for astronomers and rocketeers. He hails from the Republic of Brooklyn [and Polytechnic Institute] and now, after several trips to Himalayan mountain tops, resides in the SF Bay Area with a large book collection and several pet rocks.

Copyright © 2006, Howard Dyckoff. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

LG Tips


2-cent tip: Desktop Icon Text Color under KDE
2-cent tip: Using two computers with one kbd/mouse
2-cent tip: Using your text consoles from remote locations
2-cent tip: Recording live playback using ALSA
2-cent tip: vpac for Vim

2-cent tip: Desktop Icon Text Color under KDE

Brian Bilbrey (bilbrey at orbdesigns.com)
Sun Apr 9 14:19:33 PDT 2006

Answered by: Brian, Thomas

Google didn't help. Usually when I go to find these things, Google takes me to one of my own posts, so I can smack myself around as I gripe that "I used to know that". Even though I've put myself through this in prior iterations, I must never have posted it. Now it's time to get this information out there so that I won't be insane (over this particular issue) ever again...

Setting KDE's desktop icon text color has always, always made me insane. It's not in the same place with all the other color settings, oh, no! To set most item colors in a KDE setup, you use the Control Center, select Appearance & Themes, then Colors. A simple easy-to-navigate interface for setting almost every foreground, background and select color under KDE. But if you want to set Desktop text color ... well, that's not listed. I always end up poking and prodding for a considerable while. I found it again, today, and wanted to put it up for me and others to find. So, to set the KDE (versions 3.4 and 3.5, at least) desktop icon text, from the Control Center select Appearance & Themes. Choose Background from the sub-menu. Yeah, Background. Who knew? Then, click on the Advanced Options button. There are three sub-panes in the resulting dialog. The first, Background Program is usually all I see before bailing out of that dialog. But keep looking. Further down, there's Background Icon Text. Not Desktop, oh, no. And that's not to be found in the general color settings tab, for heaven's sake, no! Now I understand the reasoning behind where it is. But since for all the rest of the world, that's the Desktop, why not call it that, or make it easy to find using the Desktop term. And, praise Baal, why not let people change that text color in the standard tool as well as in the special secret place!

PS: Also "published" on my own site, on Monday, April 10, 2006. No worries!

[Thomas] - (Quoting Brian) Setting KDE's desktop icon text color has always, always made me insane.

Hehe. And in FVWM all that would need is one or two colorset definitions and then:

    Style * IconTitleColorset <n>, HilightIconTitleColorset <n>    

... never mind, though. :)

[[Brian]] - Yes, I know. And when I get tired and need things simplified, there's always *box (most often fluxbox). I often don't know why I keep dropping back to KDE - I think I like the kicker. I know that I use those icons on the desktop once in a blue moon, and really should just turn off "Background Icons". That would be another solution to my dilemma.

I even dabble in FVWM from time to time, and have a carefully saved FVWM configuration someplace safe (safe == well hidden). Find and locate are my friends, though.

Bottom line, however, was that finding that setting was a PITA, and I'm glad I did it, and documented it. I also bugged it as a wishlist item with the KDE team.

2-cent tip: Using two computers with one kbd/mouse

Kapil Hari Paranjape (kapil at imsc.res.in)
Sat Apr 29 20:58:37 PDT 2006

Answered by: Kapil


If some of you are unlucky(:)) enough to get confused because you have two computers on your desktop and you forget which keyboard/mouse is connected with which computer then this tip may help.

You need to choose "master" which is the computer connected to the keyboard/mouse you will actually use. You can later put away the mouse and keyboard of the other computer in order to avoid getting confused. The "master" must be an X-window machine(*) but the "slave" can be anything.

After you execute these steps and "go west" off the screen on the master your keyboard and mouse events will be directed at the slave. (**)

You need to install "x2x" and "x2vnc" on the master.

"slave" is also X-window.

   From the slave log in to the master using ssh X forwarding

		slave:$ ssh -X luser at master

	Then run "x2x" on the master via this ssh session as follows

		master:$ x2x -from :0 -to $DISPLAY -west &

	Here you replace west with the appropriate direction (west
	equals left and north equals up) in which the monitor of the
	"slave" is.

"slave" is Linux in "console" mode.

	On the slave you run "linuxvnc"

		slave:$ linuxvnc &

	This will give you a port number (usually 5900) which you
	must use below. I'll use 5900 as the port since that is
	standard. Next start an ssh tunnel to the master.

		slave:$ ssh -f -N -R 5900:localhost:5900 master 

	Finally on the master you run

		master:$ x2vnc -west localhost:0 &

"slave" is "OtherOS".

	On the slave you need a VNC server like "WinVNC" for Windows
	and "OSXVnc" for Mac OS X. You also need "ssh" unless you allow
	the VNC server to accept connections over the net (bad
	security). You then follow the same sequence:
		(a) start the VNC server on slave
		(b) start the ssh tunnel on slave
		(c) start x2vnc on master.

(*) I know that I should use the term "machine which is running an X-server" instead of X-window machine but I hope the nomenclature is clear enough. (**) There are other options like clicking in a window to switch the focus but this "trick" of going west seemed the neatest.

Hope this magic helps someone.

[Kapil] - Here is another way to do this---synergy. You can find this at


There are also Debian packages.

The simplest possible config.

On the master you create a file called (say) synergy.conf that contains:

	section: screens
	section: links
			left = lefty
			right = center

you then run:

  synergy -c synergy.conf &

On the slave you run:

	ssh -f -N -L 24800:localhost:24800 master
	synergyc -n lefty localhost

This is simpler than x2x or x2vnc and the master can also be a machine other than Linux.

2-cent tip: Using your text consoles from remote locations

Kapil Hari Paranjape (kapil at imsc.res.in)
Thu May 4 02:06:29 PDT 2006

Answered by:


You may be on an Xsession that has disabled switching to "console" mode or you may want to access the text console of a remote machine. Whatever it is two programs may be useful. One is "conspy" and the other is "linuxvnc".


Say you have access to the machine as root. Login and run "conspy 1" where you can replace 1 with the number of the
console you want to access (don't try this with the vt on which X is running!).


Suppose someone with root access on the machine is kind enough to run "linuxvnc 1" where 1 has the same meaning as above. You can then connect your favourite vncviewer to "machine:0" and access the console. http://libvncserver.sourceforge.net/

The latter has some advantages in that anyone can be given access without root priviledges. The former is distinctly "faster".

2-cent tip: Recording live playback using ALSA

Kapil Hari Paranjape (kapil at imsc.res.in)
Wed May 31 08:57:52 PDT 2006

Answered by:


Have you ever felt the urge to record some "live" music while you are listening to it? The following procedure might do what you want.

Prerequisites: The sound on your computer should use a 1.x release of the ALSA drivers and you have alsa-lib (Debian's libasound2) installed.

Assumptions: You don't already have a $HOME/.asoundrc file. If you do, then you need to edit it suitably. You also should not have a
script/program called "recording" :-)

Result: You run "recording -on" to start recording. You then start your live playback. After you have had enough you stop the live playback. You then run "recording -off" and are left with a (large) file containing the recording.


1. First create a file called "$HOME/.asoundrc" like the one enclosed.

2. Create the links.

   ln -s /dev/null /var/tmp/null.raw
   ln -s /var/tmp/null.raw /var/tmp/record.raw

3. Next install the script "recording" somewhere in your path (download it here.)

4. That's it!


a. This only works for digital live playback through the PCM device of your soundcard. You can use "arecord" for the other devices since these can usually be "captured".

b. Don't forget to copy/convert your recording /var/tmp/sound.raw to a safe location.

c. This only works with audio players that support alsa-lib. To get it to work with other players like "realplay", install alsa-oss
and use "aoss realplay" instead of "realplay".

d. You usually cannot run "recording -on/-off" while the player has the audio device open. For many players (in particular those like (c) above), you must quit the player before and after recording.

e. The supplied .asoundrc converts the incoming audio to CD format. This may lead to a loss in quality but I prefer it to "guessing" the real format of the raw data in each case.

2-cent tip: vpac for Vim

Mike Lear (mikeofthenight2003 at yahoo.com)
Tue Jun 6 03:07:02 -0700 PDT 2006

Answered by:

As a Linux and Vim user, I like to use Vim to write programs and subroutines in various programming languages. I developed 'vpac' as a series of programming aids for Perl, gcc, g++ and asm, using Vim as the focal point.

'vpac' and its related files allow you to compile, debug and run programs all within one Vim session. Another feature is the ability to comment/uncomment existing code with a single keystoke.

'vpac' works like a global makefile in that you can build several small files into one executable file. It detects the programming language that you are using and selects the appropriate compiler/assember. 'vpac' can also carry out mixed language builds, i.e. c/c++ and asm.

It is easy to use - all the commands are run from function keys.

'vpac' comes as a tar-gzipped package which includes various 'Howto's and all of the source code.


Mike Lear

Talkback: Discuss this article with The Answer Gang

Published in Issue 128 of Linux Gazette, July 2006

How Fonts Interact with the X Server and X Clients

By Thomas Adam

[ The following is taken from a reply the author sent to TAG in response to a question about fonts. Some of the answers in TAG are so good that they simply deserve to be made into articles. :) -- Ben ]


Behind the scenes, there's a fair amount that happens when an application requests the use of a font. Because many default installations include both the X server and their respective clients on the same machine, a lot of the functionality is masked. However, the X server plays a pivotal role in managing the fonts stored under it.

There are usually two different mechanisms at work, as far as fonts are concerned: one of them makes use of a font server, the other does not.

Fonts and their location

Taking a typical system that doesn't use a font server, font definitions are a property of the X server; that is, it knows and keeps track of which fonts are on your system. X11 defaults to looking for fonts in /usr/lib/X11/fonts/*.

Typically, a standard definition from /etc/X11/XF86Config-4 (and newer xorg.conf) files might look like:

Section "Files"
    FontPath        "unix/:7100"          # local font server
    # if the local font server has problems, we can fall back on these
    FontPath        "/usr/lib/X11/fonts/misc"
    FontPath        "/usr/lib/X11/fonts/cyrillic"
    FontPath        "/usr/lib/X11/fonts/75dpi/"
    FontPath        "/usr/lib/X11/fonts/100dpi/"
    FontPath        "/usr/X11R6/lib/X11/fonts/sgi"
    FontPath        "/usr/lib/X11/fonts/Type1"
    FontPath        "/usr/lib/X11/fonts/CID"
    FontPath        "/usr/lib/X11/fonts/100dpi:unscaled"
    FontPath        "/usr/lib/X11/fonts/75dpi:unscaled"

An application can request a font to display, and the X server will obligingly look for it in the hash of directories it stores (much like the one above). The command 'xset q' will list that information [1], and indeed font paths can be added to with 'xset +fp /some/location/'.

However, that does nothing more than append the directory definition. In order for the X server to become aware of the fact that a new location has been added, one has to rehash that with 'xset fp rehash'.

Font descriptions

There's a convenience mechanism within X11 fonts, and that is to alias font names. If we ignore TrueType fonts for the moment, the command 'xlsfonts' lists fonts like this:

-adobe-avant garde gothic-book-o-normal--0-0-0-0-p-0-iso8859-1
[ Many more lines elided ]

Let's take one of them as an example — here's what each part does:

font description
Figure 1: Structure of a typical font.

That's a lot of information, right? Well, yes, it is, but it's a lot of very useful information. Roughly (and off the top of my head) here's what each part means:

All of this is very dull and boring, and of course it would be a nightmare if one had to remember all of that information in one go. This is where aliasing and wildcarding become useful.

Most X11 applications that use the X11 Resource Database (XRDB) allow various resources to be set with an appropriate font. Example:

*xterm.font: *courier-bold-o-*-120*

That should be pretty self-explanatory, right? That's analogous to the often used command-line [2] of:

<program> -fn '*courier-bold-o-*-120*'

The X server then has to look up that font, expanding the wildcard as it goes. This is largely left down to the user to ensure that the correct placement of any wildcards is accurate, since it will on many occasions match nothing or unintentional fonts due to it. The X server will traverse whatever is in its fontpath, in the order the directories are listed, until a matched font is found. I cannot stress the order enough — it's analogous to the way a binary is searched for, in one's $PATH. The first matching font is whatever is found within the list of fontpaths, even if two or more fonts are matched by the wildcard.

Aliasing is slightly different, in that rather than the user relying on wildcard matching, a "font.alias" file holds short names for fonts (alternate names, if you will). Here's a snippet of one:

lucidasans-bolditalic-8    -b&h-lucida-bold-i-normal-sans-11-80-100-100-p-69-iso885 9-1

Essentially, it's a two column file, with the alternate name in column one, and the actual font name in the second column. As before, if you use an alias to load a font, the X server will search each font directory in turn. This has the added benefit of being able to specify aliases for fonts in other directories.

A fonts.alias file is also associated with a fonts.dir file. You can think of this file as a massive database that the XServer uses. It's a bit like font.alias, except that this file lists the following:

Actual Font Name		Font Name

When you ask the X server to search for a font, it will look in fonts.dir to ascertain the font based on either the long name, or the alias (since the alias is mapped before the fonts.dir is looked in.) If you've ever used the mkfontdir(1) command, this is what it does — creates font.dir files in each and every fontpath listed.

Font Servers

Now onto font servers: You don't need them. Really — unless you're in some large multinational corporation that has hundreds of workstations connecting to an X server with different vendors. In the R5 release of X11, they were used for uniformity, to ensure that font names remained consistent, so that applications could load fonts, thus sharing them. What happens is something like the following:

XFS Comms
Figure 2: How XFS interacts with the X server and X client(s)

The machine "Server" has a number of services running on it -- including the XFS (X Font Server). The local X server running on a client is hence told to use a font server (which is typical of the line):

FontPath "unix/:7100" 

The font server responds by supplying the X server on the client with a list of font names applications (X clients) can load and display on the screen. (Under the hood there's a lot which goes on, but I'll skip that.) Note the "role reversal" here [3]: the X server is the client with respect to the font server — hence it is itself a "font client" — other examples include a printer, which would also talk to the font server, where necessary (although not shown in the above diagram).

Old versions of Red Hat used to insist on running a TrueType font server, for no other reason than presumably to annoy everyone.

[1] Programatically, this can be achieved via the XSetFontPath() call.

[2] Note the hard quotation marks here, so as not to perform globbing at the shell.

[3] Rick Moen comments: Unix newcomers are often confused by the notion of applications being X11 clients and the graphical display being driven by a server process, which somehow is opposite to people's expectations. They think: surely the applications are serving up display data, thus making them servers, which the graphical display is receiving, thus making it a client. However, what's being served up, to both local applications and remote ones via X11 network access, is the graphical display software's drawing services, as a central system facility for all applications that need it. Thus the applications are, for that purpose, clients.

Talkback: Discuss this article with The Answer Gang


I used to write the long-running series "The Linux Weekend Mechanic", which was started by John Fisk (the founder of Linux Gazette) in 1996 and continued until 1998. Articles in that format have been intermittent, but might still continue in the future. I currently write occasional articles for LG, whilst doing a few things behind the scenes for it. I'm also a member of The Answer Gang.

I was born in Hammersmith (London UK) in 1983. When I was 13, I moved to the sleepy, thatched-roofed, village of East Chaldon in the county of Dorset. It is very near the coast, and Lulworth Cove, which is where I used to work. Since then I have moved to Southampton, and currently attend University there, studying for a degree in Software Engineering.

I first got interested in Linux in 1996 having seen a review of it in a magazine (Slackware 2.0). I was fed up with the instability that the then-new operating system Win95 had and so I decided to give it a go. Slackware 2.0 was great. I have been a massive Linux enthusiast ever since. I ended up with running SuSE on both my desktop and laptop computers. Although I now use Debian as my primary operating system.

I am actively involved with the FVWM project, writing documentation, providing user-support, writing ad-hoc and somewhat esoteric patches for it.

Other hobbies include reading. I especially enjoy reading plays (Henrik Ibsen, Chekhov, George Bernard Shaw), and I also enjoy literature (Edgar Allan Poe, Charles Dickens, Jane Austen to name but a few).

I am also a keen musician. I play the piano in my spare time.

Some would consider me an arctophile (teddy bear collector).

I listen to a variety of music.

Copyright © 2006, Thomas Adam. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

Creating a Rudimentary Kiosk System using FVWM

By Thomas Adam


It is always interesting to see how and where Linux is used within the wider community, especially where the general public are concerned. Indeed, it is quite often the case for public libraries to use a batch of computer terminals to offer a portal via their own intranet so that they can do searches for books, renew overdue books, etc.

However, ensuring that those terminals only provide that specific service can be something of a challenge. Because they're public-facing terminals, their functionality is likely to be limited to a specific subset of the available applications. Ensuring that these applications are the only ones the user of the terminal sees, along with making sure that no other application is accessible unless it's explicitly needed in such a way that it doesn't break the system can be an interesting task, to say the least.

This is where some sort of kiosk comes into play. Note that the concept is nothing new, although the current HOWTO on the issue is somewhat out of date. FVWM, which is mentioned in the above HOWTO, has advanced a lot since it was written. Of course, since then, desktop environments such as KDE have also had a kiosk extension added to them.

This article will focus on using Firefox as the main and only application the user will be using. By using FVWM to manage this application, it is possible to provide restrictions not only on the application, but also the environment the user will be using.

The motivation for this article is mostly derived from a series of e-mails I have been exchanging with someone who seems to be trying to setup a kiosk system in a corporate environment. The actual content of the e-mail that turned into this article (as a very odd coincidence) is born out of an e-mail from another TAG member, who asked the same question.

Originally, I had planned to discuss how to lock down Firefox. While this is applicable within the context of this article, it's also somewhat less restrictive, given that different people require different circumstances. Firefox itself has a series of extensions dedicated to kiosk browsing that you can further use to lock Firefox down. This article will concentrate on getting a window manager (FVWM) to do the rest of the work.


In the best case, what you'll probably want to do is something like the following:

From first principles, here's what your ~/.fvwm/config file ought to contain. Note that any decoration changes for this are out of scope for this article; if you don't like hideous pink borders, change them. (I'm going to suggest FVWM version >=2.5.16; anything I write here won't work on FVWM 2.4.X).

Enter FVWM

StartFunction is a function that FVWM runs at both restarts and init. So, in this, all that's needed is to start your application. Since there are some crude preventative measures one can take to ensure this application doesn't die once the WM has loaded, we'll just start this at init time.

DestroyFunc StartFunction
AddToFunc   StartFunction
+ I Test (Init) Exec exec firefox

That's that, all taken care of. Next thing to consider is whether you want any window decorations or not, for Firefox. Let's assume for the moment that you do want the title and borders, but want to restrict the buttons that appear on the title. That's fine; just set a style line for it:

Style Firefox-bin NoButton 1, NoButton 2, NoButton 4, ResizeHintOverride

NoButton removes a button from being displayed in the title. ResizeHintOverride makes those column-dependent aware applications not worry about it. The number buttons are ordered thus, on a titlebar:

| .  .  .  .  .                              .  .  .  .  . |
| 1  3  5  7  9                             10  8  6  4  2 |
|                                                          |

Typically (and as is the default), buttons 1, 2, and 4 are defined for normal operations of displaying a menu (button 1) and minimising and maximising an application (buttons 4 and 2 respectively).

The next thing to do is remove any possible mouse bindings on the title bar that could disrupt its operation, or that could cause some arbitrary operation to be performed on it. This also includes bindings for both the frame and the sides of the window.

Mouse 1 SF A -
Mouse 1 T  A -
Mouse 2 SF A -
Mouse 2 T  A -

That's just an example. You ought to do the same for key bindings, although by default FVWM defines no key bindings other than bringing up a popup menu.

So what happens once the application starts? You'll presumably want it maximised. Read the FvwmEvent article for details of this. Essentially, the following could be used:

DestroyModuleConfig FE-SM: *
*FE-SM: Cmd Function
*FE-SM: add_window StartFirefoxMaximised

Module FvwmEvent FE-SM

That sets up FvwmEvent using a module alias of 'FE-SM'. It's listening for the add_window event, and hence will call the 'StartFirefoxMaximised' when any new window is mapped. As for what StartFirefoxMaximised looks like:

DestroyFunc StartFirefoxMaximised
AddToFunc   StartFirefoxMaximised
+ I ThisWindow (Firefox-bin, !Maximized) Maximize 

The function says that if the window mapped matches "Firefox-bin" as either its window name, class, or resource (class in this case), and it is not maximised, to maximise it.

And that's it, right? Pffft, if only. What if, by some means unknown to us, the window were un-maximised? That's not something we probably want within this kiosk environment, much less allowing the window to be moved once it has been maximised. That's OK, since we can restrict this. You might think it's a simple case of adding to the style definition we defined for Firefox earlier:

 Style firefox  NoButton 1, NoButton 2, NoButton 4, FixedSize, FixedPosition, !Maximizable

But that's not quite true. What happens here is that the style preferences are applied before the add_window event is triggered. Whilst it is true that "FixedSize" and "FixedPosition" do exactly what we want (i.e., doesn't allow resizing, or moving the window, or not allowing the window to become un-maximized, respectively), we have to apply it afterwards. This is where the WindowStyle command can be used.

The WindowStyle command works like Style, except that it pertains to a specific window currently mapped and visible. Under the hood, it just assigns various struct events via the window's WindowId, but that's out of scope, here. Hence in the StartFirefoxMaximised function, we can now expand upon this further:

DestroyFunc StartFirefoxMaximised
AddToFunc   StartFirefoxMaximised
+ I ThisWindow (Firefox-bin, !Maximized) Maximize 
+ I ThisWindow WindowStyle FixedSize, FixedPosition, !Maximizable
+ I UpdateStyles

I mentioned earlier that you may or may not want any window decorations. If that's the case, then you can do the following to turn them off:

Style firefox !Title, !Borders, HandleWidth 0, BorderWidth 0

That's a bit better, right? Well, maybe. There's still a few other considerations that should be taken into account. When a window is mapped, it is put into a layer (layer 4 by default, which can be changed using the DefaultLayers command). This is fine, but means some windows can cover up this window. This is unlikely to happen here, given that it's unlikely you're going to be running any other application, but, for the theoretical concept alone, it's best to perhaps put this window into a much higher layer than normal.

Style Firefox-bin  NoButton 1, NoButton 2, NoButton 4, Layer 8

FVWM (like several other window managers) has virtual desktops. Wooo. You aren't going to need them here, so ditch them. You can do this using the DesktopSize command, and perhaps the DesktopName command , to define how many desks you want.

DesktopSize 1x1
DesktopName Main

Hence the DesktopSize command restricts us to one page defining a single desk, whose name is "Main". I also alluded to earlier ensuring that you turn off all mouse bindings. This is also true of the root window that might bring up any menus.

Mouse 1 R A -
Mouse 0 R A -
Mouse 2 R A -

Again, it's unlikely to much of an issue, given the application runs in some sort of full screen mode, but it's better to cover one's bases than not at all.

There are two more things that need mentioning. Ensuring that the Firefox window is the only running instance can be a little tricky. We could prepend some stuff to the StartFirefoxMaximised function to try and enforce such a policy.

DestroyFunc StartFirefoxMaximised
AddToFunc   StartFirefoxMaximised
+ I ThisWindow (!Firefox-bin) Close
+ I ThisWindow (Firefox-bin, !Maximized, !Transient) Maximize 
+ I TestRc (Match) ThisWindow WindowStyle FixedSize, FixedPosition, !Maximizable
+ I UpdateStyles

...although that may or may not be overkill for some purposes. It might be the case, of course, that you only want to allow a certain subset of applications to run. This would presumably be via Firefox itself (perhaps some filetypes spawn mplayer or RealPlayer), hence you could expand upon StartFirefoxMaximised even more.

DestroyFunc StartFirefoxMaximised
AddToFunc   StartFirefoxMaximised
+ I ThisWindow (!"App1|App2|App3|App4") Close
+ I ThisWindow (Firefox-bin, !Maximized) Maximize 
+ I ThisWindow WindowStyle FixedSize, FixedPosition, !Maximizable
+ I UpdateStyles

'ThisWindow (!"App1|App2|App3|App4") Close' says something like the following: "If the window just created is not one of App1 or App2 or App3 or App4, then close it." Thus, in this way, one can conditionally place restrictions.

One final consideration. How do you ensure that the Firefox window is never closed? Well, FVWM has a style consideration "!Closable" that when applied means the application cannot be closed... almost. It only goes as far as trying to circumvent the various events generated. It does not, for example, stop xkill closing the application, or for some external source to do so.

You could use some sort of script to monitor whether the application is still running (which assumes this script also runs Firefox).

firefox &          

while sleep 60; do
  [ kill -0 $ffpid ] || {
	# Spawn firefox
	firefox &

But that's far from ideal. You might be better off using FvwmEvent to monitor when windows are closed, and to respawn them as apropos. Hence adding to the "FE-SM" definition from earlier.

*FE-SM: destroy_window CheckWindowClosed 

Tells that module alias to now also listen for any windows closing and take action. The function CheckWindowClosed might look like this.

DestroyFunc CheckWindowClosed
AddToFunc   CheckWindowClosed
+ I ThisWindow (Firefox-bin) Exec exec firefox

That works, right? Well, yes it does. But what if there were more than one instance of Firefox running? What if we wanted to respawn this application only if the last remaining instance of this application died? That's no problem:

DestroyFunc CheckWindowClosed
AddToFunc   CheckWindowClosed
+ I None (Firefox-bin) Exec exec firefox

None will only work if there's no instances of the matched window being asked for — in this case, "Firefox-bin".


Some may argue that the use of a window manager is in itself overkill, given that one could easily dispense with using one, and just start Firefox in fullscreen mode as soon as X11 is logged into. However, that's potentially disastrous, given that the windows aren't controlled. A transient window (such as an open dialogue box) wouldn't be managed — if it were opened in an awkward place on the screen, there's be no way of moving it. Neither, of course, does not using a window manager preclude the possibility of closing other windows (applications) down that don't ultimately belong there.

In short, there is no real solution or easy answer to creating a true kiosk system. At best it can simply be managed by enforcing certain rules specific to its environment.

Talkback: Discuss this article with The Answer Gang


I used to write the long-running series "The Linux Weekend Mechanic", which was started by John Fisk (the founder of Linux Gazette) in 1996 and continued until 1998. Articles in that format have been intermittent, but might still continue in the future. I currently write occasional articles for LG, whilst doing a few things behind the scenes for it. I'm also a member of The Answer Gang.

I was born in Hammersmith (London UK) in 1983. When I was 13, I moved to the sleepy, thatched-roofed, village of East Chaldon in the county of Dorset. It is very near the coast, and Lulworth Cove, which is where I used to work. Since then I have moved to Southampton, and currently attend University there, studying for a degree in Software Engineering.

I first got interested in Linux in 1996 having seen a review of it in a magazine (Slackware 2.0). I was fed up with the instability that the then-new operating system Win95 had and so I decided to give it a go. Slackware 2.0 was great. I have been a massive Linux enthusiast ever since. I ended up with running SuSE on both my desktop and laptop computers. Although I now use Debian as my primary operating system.

I am actively involved with the FVWM project, writing documentation, providing user-support, writing ad-hoc and somewhat esoteric patches for it.

Other hobbies include reading. I especially enjoy reading plays (Henrik Ibsen, Chekhov, George Bernard Shaw), and I also enjoy literature (Edgar Allan Poe, Charles Dickens, Jane Austen to name but a few).

I am also a keen musician. I play the piano in my spare time.

Some would consider me an arctophile (teddy bear collector).

I listen to a variety of music.

Copyright © 2006, Thomas Adam. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

A Brief Introduction to VMware Player

By Edgar Howell

The other day on the way home from work, between the streetcar and the bus, I spent a couple of minutes at the train station bookstore, and picked up a magazine with a DVD containing 10 new "Linux-PCs". Hmmmm...

After getting off the bus, but before I had a chance to complete the walk home, it began to rain. That ruined all hopes of being able to mow the lawn... so I decided to investigate the DVD, instead.

As it turned out, it contained VMware Player and 10 virtual machines, or VMs: Damn Small Linux, Fedora Core 4, Pocketlinux, and so on. It was beginning to look as if the weekend would have some sunshine, after all!

Just What Is VMware?

For those not familiar with virtualization in general, or VMware in particular, virtualization has been around in the mainframe world for some 30-odd years. It enables effectively simultaneous execution of more than one instance of operating systems — including, in the case of VMware, ones that would not normally even be able to execute on the physically available hardware. This is a remarkable accomplishment that requires convincing the operating system that it really has sole ownership of the hardware environment it requires.

This unfairly terse description falls horribly short of doing justice to the amazing feat - but there is no shortage of far more detailed descriptions on the Internet.


As usual, the first step is to install VMware Player. Well organized by VMware, installation was a real no-brainer. I did read the prompts and consider the responses, but wound up just accepting the defaults offered by the installation script.

Perhaps it was an idiosyncrasy of the particular magazine that I picked up (it actually addresses users of Wimp/OS), but they had used a compression program I neither had nor had ever heard of. No problem - it was available at SourceForge, and just needed to be installed.

[ '7zip' is actually not a Wind0ws-centric program at all; it's been available for Linux for quite a while now. To quote Debian's package listing, "7-Zip is the file archiver that archives with the highest compression ratios. The program supports 7z (that implements LZMA compression algorithm), ZIP, Zip64, CAB, RAR, ARJ, GZIP, BZIP2, TAR, CPIO, RPM and DEB formats. Compression ratios with the new 7z format are 30 to 50% better than ratios with the ZIP format." Good stuff, in other words... except that their RAR compressor is non-free. -- Ben ]

web@linux:~/vm> cp /media/usbdisk_1/p7zip_4.39_x86_linux_bin.tar.bz2 .
web@linux:~/vm> ls
web@linux:~/vm> bunzip2 p7zip_4.39_x86_linux_bin.tar.bz2
web@linux:~/vm> ls
web@linux:~/vm> tar xf p7zip_4.39_x86_linux_bin.tar
web@linux:~/vm> ls p7zip_4.39
bin  ChangeLog  contrib  DOCS  install.sh  man1  README  TODO
web@linux:~/vm> ls p7zip_4.39/bin
7z  7za  7zCon.sfx  7zr  Codecs  Formats
web@linux:~/vm> cp p7zip_4.39/bin/7za .
web@linux:~/vm> l
total 8360
drwxr-xr-x   3 web users    4096 2006-05-20 17:24 ./
drwxr-xr-x  15 web users    4096 2006-05-20 17:20 ../
-rwxr-xr-x   1 web users 1198328 2006-05-20 17:24 7za*
drwxr-xr-x   6 web users    4096 2006-05-20 17:24 p7zip_4.39/
-rwxr-xr-x   1 web users 7331840 2006-05-20 17:23 p7zip_4.39_x86_linux_bin.tar*

After that, installation of the VMs was quite straight-forward.

web@linux:~/vm> cp /media/dvd/ubuntu/computer/pc_gratis/dsl-linux.exe .
web@linux:~/vm> ./7za x dsl-linux.exe

7-Zip (A) 4.39 beta  Copyright (c) 1999-2006 Igor Pavlov  2006-04-13
p7zip Version 4.39 (locale=en_US.UTF-8,Utf16=on,HugeFiles=on,1 CPU)

Processing archive: dsl-linux.exe

Extracting  dsl-linux
Extracting  dsl-linux/nvram
Extracting  dsl-linux/dsl-test-250.vmdk
Extracting  dsl-linux/other24xlinux.vmx

Everything is Ok

Now, VMware Player was fully functional and could execute the available VMs.

Using Virtual Machines

Under SUSE with KDE, click on the panel icon for K Menu, then 'System' -> 'More Programs' -> 'VMware Player'.


Navigation is not particularly challenging. As a first test, I chose Damn Small Linux from the 'vm' directory.

Solitaire is a bit sluggish on this 2.6 GHz Celeron(tm), but bearable. In any case, the following definitely isn't SUSE!


When the bottom line shows "To grab input, press Ctrl+G", you can move the cursor over the window as usual. Once you press Ctrl+G, the cursor belongs to that window, and it is necessary to press Ctrl+Alt to effectively de-activate that window and free the cursor. Ctrl-Alt-F2 and the like still work as usual.

At this point, I halted the VM and shut VMware Player down. Then, I rebooted the hardware. Upon restarting this VM, the former status was restored:


Before re-starting VMware Player, I also added another couple of VMs. The following looks interesting for this long-time SUSE user, at least:


Indeed, after logon:


And for the skeptics in the crowd, yes, both are active at the same time:


The initial boot of a VM takes quite a while, just as does booting hardware. On the other hand, restarting a VM that has been active for a while is extremely fast.

Activating the network was no problem at all. In each of two VMs, after a right-click on the "Ethernet" button, I selected "host-only" and then assigned IP-addresses as in "ifconfig eth0 up". After that, ping worked as expected:


Sharing Hardware

VMware Player seems to share the hardware with the native operating system quite well. While using Midnight Commander under SUSE to copy VMs from the DVD into the appropriate directory, I started Ubuntu under VMware, and it discovered the DVD drive and went into update mode (synaptic). That slowed down the copying considerably, but it worked just fine.

It is interesting that there is almost no limit to the number of VMs available. VMware doesn't use partitions; each VM exists within its own directory. I renamed the Ubuntu directory and unpacked it again, creating a clone. When the renamed VM was started, it came up with the status at the point where it had been shutdown. The new VM went through the usual (simulated) hardware boot.

The SUSE VM is 10.0, which is my most-current version. After placing the distribution DVD in the drive and clicking on the "CD-ROM" button, there was no problem using YaST to add software to the VM. Although VMware Player doesn't support creating new VMs, they clearly can be updated. Since YaST uses RPMs, installing one ought to work, and I would expect that to be the case for a tarball, as well.

Finite Resources

But don't get over-confident in starting VMs! At least my resources are finite:



After assigning an appropriate IP-address such that the Ubuntu VM was on the same sub-net as the network printer, it was possible to ping it. Similarly for the notebook, manually assigned IP-address This required using the "Ethernet" button to change the status of the network from "host-only" to "bridged".


There was no trouble using SSH to log onto the notebook from Ubuntu, or the other way around, although this of course required using the ad-hoc IP-addresses rather than host names.

Why Bother?

Some time ago, when I mentioned to someone that I was interested in checking out virtualization on a PC, his response was "Why?" Wow, someone even more skeptical than I am! But the question is certainly legitimate.

Typically the "traditional" reason to use virtualization in this environment has been consolidation of servers, as in a server farm that has lots of hardware idle most of the time. Particularly if hosting for customers who may be competitors, virtualization provides clean separation of VMs as opposed to simply providing service for many customers on one machine.

However, I don't have a server farm in need of consolidation in the office at home, and I doubt that most readers do either. However, if you think about it, this can be a very interesting technology, although certainly not of use to everybody.

As you saw above, given available VMs, it is essentially trivial to make an unfamiliar environment available in a non-trivial way. As a long-time SUSE user, I'm not at all familiar with GNOME. Disparaging comments by others notwithstanding, I am interested in playing around with GNOME, which is very different from what I am used to.

Indeed, this environment should be ideal for both teachers and students. One of the things that I really appreciate about GNU/Linux is the ability to learn new things at my own pace, at home, when I have the time and at no or minimal expense. Without a great deal of effort on my part, I can now get familiar with several distributions I might never have had the time for.

Someone teaching a course could set up a series of exercises for his students, and just needs to copy a handful of VMs onto CD-ROM. This would be very effective for a course on some Linux distribution, particularly if the students have only Wimp/OS at home.

Or consider the need to do regression tests. The fact that every VM restarts at the point it had been shut down would be extremely nice for testing. Set up initial conditions and save the directory to ensure that every test later run actually is testing the same thing, unaffected by any side-effects of something else run between tests.

Philosophy 101

To those who might object to VMware's proprietary business model, I would reply that it is another tool on the market, and, from my perspective as a consultant, I like having it in my toolbox. Very many years ago, on a job-interview, I had to admit that I didn't yet have any experience with one particular package. Someone else got that job. Made sense then, makes sense today. Do you want to be the one to say "I can learn that"? How about "Been there, done that"?

In my opinion, virtualization is going to be very important, very soon. I am really looking forward to my first machine with Vanderpool or Pacifica technology.


This "brief" introduction comes nowhere near doing justice to the accomplishment of VMware. It can be installed under Wimp/OS to run GNU/Linux or the other way around (if your proprietary EULA license gives you access to or permits your producing the appropriate VM). Or merely to isolate a test-bed from a production environment, as in the above.

There were problems, but not all of them were due to the product. Likely attributable to VMware are problems with reboot and date/time, although it might be argued that these are not bugs but features. Using "reboot" or "shutdown -r now" returns the VM to single-user mode. I couldn't figure out how to get "startx --:1" under SUSE to do any good. And after having used VMware's "Quit" button to shut down a VM for a couple of days, on restart it has the date and time of shutdown. Well, in some scenarios that does make sense.

VMware certainly does require significant resources. The version of Solitaire that comes with SUSE will finish a game automatically if told to try; natively, on a notebook with 1 GB of memory, it moves about 5 cards per second. Under VMware, however, it takes about 2 seconds per card. ping can take so long that it seems to have hung: in spite of reporting 0.352 ms, the wall-clock time was closer to 10 seconds!

This is not the responsibility of VMware, but my PC has only 512 MB - which sure seemed like a lot at the time. Two instances of Ubuntu can be started in that, but their response times are egregiously slow. A third instance fails due to lack of memory.

A general problem with an environment where several different systems can be played with is inexperience with each, by definition (again, not relevant to VMware specifically.) It took frustratingly long to find a command line with GNOME. Although Ubuntu claimed to be using a German keyboard, it wasn't and I couldn't convince it to do so — but, then, SUSE with the same problem was difficult, because SaX2 wouldn't execute consistently.

If the VM available to you doesn't correspond to your hardware, you might be in trouble. Given appropriate distribution media, it is possible to update software. However, what if the VM doesn't even know about DVDs or writing to CD? Well, there's always networking out to the native operating system.

Nonetheless, never was it so easy -- without having to worry about repartitioning or zapping the MBR -- to get so many different operating systems functional on one machine, not only without their getting in each other's way but able to co-operate with each other.

Talkback: Discuss this article with The Answer Gang

Bio picture Edgar is a consultant in the Cologne/Bonn area in Germany. His day job involves helping a customer with payroll, maintaining ancient IBM Assembler programs, some occasional COBOL, and otherwise using QMF, PL/1 and DB/2 under MVS.

(Note: mail that does not contain "linuxgazette" in the subject will be rejected.)

Copyright © 2006, Edgar Howell. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

Subversion: Installation, Configuration — Tips and Tricks

By Muthaiah Ramanathan


This article is about installing and configuring Subversion v. 1.3.x; the installation was performed under Fedora Core 4, but should not be very different under other distributions. This article will not touch upon the usual subjects — i.e., where to download Subversion and what distro-specific commands should be used to install and configure this open source version control system.

To the novice/beginner, Subversion is a version control tool from the free/open-source world, designed to replace (and completely take over) the old horse, CVS. We will not be delving in-depth into how and why Subversion works the way it is working currently, and that is not the objective of this article.

Multiple Subversion(s)

Using system commands, check whether you already have Subversion installed. If so, verify the version before proceeding to install the latest. It is quite possible to have multiple versions installed in the same system.

Multiple Subversions — v. 1.2.3 and v. 1.3.0 — are installed in this system, for example:

[ram@lemuria ~]$ whereis svnversion
svnversion: /usr/bin/svnversion /usr/local/bin/svnversion /usr/share/man/man1/svnversion.1.gz
[ram@lemuria ~]$ /usr/bin/svnversion --version
svnversion, version 1.2.3 (r15833)
   compiled Aug 26 2005, 03:42:45

Copyright (C) 2000-2005 CollabNet.
Subversion is open source software, see http://subversion.tigris.org/
This product includes software developed by CollabNet (http://www.Collab.Net/).

[ram@lemuria ~]$ /usr/local/bin/svnversion --version
svnversion, version 1.3.0 (r17949)
   compiled May 20 2006, 23:55:41

Copyright (C) 2000-2005 CollabNet.
Subversion is open source software, see http://subversion.tigris.org/
This product includes software developed by CollabNet (http://www.Collab.Net/).

Pre-Installation Checks

Subversion repository storage can be based on any one of the following:

At the time of creating the repository, the user has the choice of using either one of the storage types to hold the repository. By default, the storage type is FSFS enabled. [1]

With Subversion, access to the repository also has multiple choices, which can be any one among the following:

  1. local file-system access,
  2. HTTP-based access, and,
  3. secure shell-based access.

How does one know for which mode of "repository access" Subversion has been configured/installed? To see the answer, run the following command:

svn --version

On my system, the output from the above command looks like this:

[ram@localhost ~]$ svn --version
svn, version 1.3.1 (r19032)
   compiled Apr  4 2006, 06:38:16

Copyright (C) 2000-2006 CollabNet.
Subversion is open source software, see http://subversion.tigris.org/
This product includes software developed by CollabNet (http://www.Collab.Net/).

The following repository access (RA) modules are available:

* ra_dav : Module for accessing a repository via WebDAV (DeltaV) protocol.
  - handles 'http' scheme
  - handles 'https' scheme
* ra_svn : Module for accessing a repository using the svn network protocol.
  - handles 'svn' scheme
* ra_local : Module for accessing a repository on local disk.
  - handles 'file' scheme

Q&A: Before Installing Subversion

Take time and try to find the answers to the following questions before proceeding to install and configure Subversion.

Whatever your answers, always create one special/common Subversion group ("svnusers") to which users can be assigned; to these users, Subversion repositories can be opened-up for access. (For example, with the common group, the repository could be a play area/scratch pad sort of setup, where "everyone" can practise Subversion and familiarise themselves with the tool.)

The next task is to create a special user ("apache") for running the "httpd" server. This user can be added to that special/common group created for the needs of Subversion.

On my system, the results look like this:

[ram@localhost ~]$ grep apache /etc/passwd
[ram@localhost ~]$ grep svn /etc/group

NOTE: The "apache" user is part of the "svnusers" group, but can be configured in such a way as to have two special groups for Subversion — one group for those that access the repositories via "http" and another for access via "file-system".

One final word on Web server "httpd". Always install the "httpd" server first, before Subversion. Subversion is built on APR (Apache Portable Runtime) libraries, which provide all the interfaces that Subversion requires for its basic functionalities on different OSes, like disk access, network access, memory management, etc.

Subversion with Apache

Pre-Installation Tips

It is always good to install Subversion from its sources. [2] In such a case, the "configure" script from the Subversion source should be run like this (assuming that the "httpd" server is already installed):

./configure --with-apr=/full/path/to/apr-config --with-apr-util=/full/path/to/apr-util-config

Post-installation Tips

After installing Subversion, I proceeded to play around with it using "http" access, but things did not happen as I expected: A quick look at the Web server logs revealed that "http" requests were failing. Thanks to people on the 'svnforum' mailing list, the problem was quickly revealed: it concerned the APR libraries used in Subversion source compilation.

Using the "ldd" command, I verified the APR libraries used in 'mod_dav_svn.so' and 'httpd', but they did not match — meaning that "httpd" and mod_dav_svn were not referring to the same APR libraries they should be.

[ram@lemuria ~]$ ldd /lnx_data/apache2.0.54/modules/mod_dav_svn.so | grep apr
        libaprutil-1.so.0 => /lnx_data/apache2.0.54/lib/libaprutil-1.so.0 (0xb7def000)
        libapr-1.so.0 => /lnx_data/apache2.0.54/lib/libapr-1.so.0 (0xb7d59000)
[ram@lemuria ~]$ ldd /lnx_data/apache2.0.54/bin/httpd | grep apr
        libaprutil-1.so.0 => /lnx_data/apache2.0.54/lib/libaprutil-1.so.0 (0xb7f8f000)
        libapr-1.so.0 => /lnx_data/apache2.0.54/lib/libapr-1.so.0 (0xb7f6d000)

"mod_dav_svn.so" should be linked to the same APR libraries as used by "httpd", and fixing that took care of the problem.

Repository Access using HTTP

Following successful installation of Apache and Subversion, next, repository access via http:// schema must be configured in "</path/to/apache/conf>/httpd.conf". Some of the parameters that can be configured are related to hosting the repository; other parameters are related to the special/common user and group for Apache and Subversion:

# If you wish httpd to run as a different user or group, you must run
# httpd as root initially and it will switch.
# User/Group: The name (or #number) of the user/group to run httpd as.
# It is usually good practice to create a dedicated user and group for
# running httpd, as with most system services.
User apache
Group apache_svn

After creating the repository and configuration, it's time to bring-up the Web server and start accessing the repository via http:// schema. As you can see below, the "httpd" server is started and running as that special/common user, "apache":

[ram@lemuria ~]$ ps auxw | grep apache
root      2984  0.7  0.6   7616  2848 ?   Ss   19:46   0:00 /lnx_data/apache2.0.54/bin/httpd -k start
apache    2985  0.0  0.4   7616  1900 ?   S    19:46   0:00 /lnx_data/apache2.0.54/bin/httpd -k start
apache    2986  0.0  0.4   7616  1900 ?   S    19:46   0:00 /lnx_data/apache2.0.54/bin/httpd -k start
apache    2987  0.0  0.4   7616  1900 ?   S    19:46   0:00 /lnx_data/apache2.0.54/bin/httpd -k start
apache    2988  0.0  0.4   7616  1900 ?   S    19:46   0:00 /lnx_data/apache2.0.54/bin/httpd -k start
apache    2989  0.0  0.4   7616  1900 ?   S    19:46   0:00 /lnx_data/apache2.0.54/bin/httpd -k start

Configuring svnserve

As stated in the Subversion manual, the svnserve program is a lightweight server capable of communicating to clients over TCP/IP using a custom, stateful protocol. Clients can use svnserve services using either svn:// or svn+ssh:// access schema. However, before that, svnserve must be configured. Here, again, there are multiple options: svnserve can be configured to run by "inetd" or as a standalone daemon process.

Authentication Options

In the following examples I have configured two separate "svnserve" daemons, again on the same system, both serving different repositories and listening to different ports. For one of the repositories, authentication options are turned on using Subversion's built-in authentication and authorization mechanisms.

The result of configuring multiple repositories under "svnserve" is shown here.

Subversion's built-in authentication configured using "/path/to/repository/conf/svnserve.conf" is shown here.

The svnserve.conf shown in the previous example has many entries and one among them is password-db = passwd. This passwd file can be located in the same directory as "svnserve.conf" and sample entries of this file are shown here.

If "passwd-db = passwd" is turned on but there are no real entries in the "passwd" file, you'll see errors like this one:

[ram@lemuria svntest_project]$ svn commit -m "added the first test C file"
svn: Commit failed (details follow):
svn: /lnx_data/report_holder/svntest_project/conf/svnserve.conf:18: Section header expected

As can be seen here, commit operations will run successfully once the appropriate entries are defined in svnserve.conf and passwd.

NOTE: The passwd file referred to above is not the same as the system file, /etc/passwd. The system file cannot be used in svnserve's configuration.


In this brief article, I have tried to highlight basic parameters and options usable for configuring Subversion with or without Apache. It's also a good idea to go through Subversion documentation to clarify your doubts on this tool's usage. Many a time, it is worthwhile to consult people (e.g., the Subversion users mailing list) and seek guidance towards resolving Subversion problems.

[1] Rick Moen comments: Since all Linux Gazette editorial production is managed via svn (Subversion), we've tallied up considerable experience with both of these repository options over the years, and can heartily recommend the stability advantages of the newer FSFS option, which uses platform-independent flatfiles rather than database storage: Our experience with the older BerkeleyDB ("BDB") option was that improperly terminating svn processes would often wedge the repository if they died during commits, requiring an irksome "recovery" operation before work could resume. FSFS repositories are also less trouble to backup.

[2] Rick Moen comments: The usual cranky-sysadmin advice applies: In my view, as a matter of administrative best practices, compiling source code from an upstream maintainer's source tarball should be your last resort, if good official packages from your Linux distribution or trustworthy, quality unofficial packages constructed for your Linux distribution don't exist, only then should you use upstream tarballs as a fallback. Even then, check to see if the the tarball provides build instructions, e.g., an rpm SPEC file, for your distribution.

Why? Any time you skip the work done by distribution package maintainers and work directly with upstream source code, you're implicitly agreeing to either perform all of that work yourself or suffer the consequences of its absence. Most people have little conception of how much work that is, or its value. Package maintainers monitor upstream releases to select which have the best mix of new features and debugging (newer not necessarily being better), usually apply patches to adapt them to distribution-specific needs, and sometimes code and apply security patches the upstream programmer never even thought of. Moreover, they do this on an ongoing basis, making improvements available to a myriad of systems, keeping those systems updated without their admins needing to worry, let alone duplicate that work.

Talkback: Discuss this article with The Answer Gang

Bio picture

I hold an engineering degree in Electronics & Communication.

After finishing my degree in '95, I worked in customer support, system administration, and lab/tools administration in various start-ups and in Cisco's offshore development center in Chennai, India.

Currently, I am working at Infineon Tech India in the SCM (software config mgmt) domain, leading a team of 6 engineers specializing in IBM Rational ClearCase and ClearQuest with experience in Shell and Perl scripting.

Copyright © 2006, Muthaiah Ramanathan. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

Coding a Simple Packet Sniffer

By Amit Saha

This article has been removed from this issue following revelation of excessive direct copying from a primary source. Linux Gazette was not aware of that copying until recently, and regrets it.
-- Ben Okopnik, Editor-in-Chief

Talkback: Discuss this article with The Answer Gang

Bio picture

The author is a 3rd year Computer Engineering Undergraduate at Haldia Institute of Technology, Haldia. His interests include Network Protocols, Network Security, Operating systems, and Microprocessors. He is a Linux fan and loves to hack the Linux kernel.

Copyright © 2006, Amit Saha. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006

The Foolish Things We Do With Our Computers

By Ben Okopnik

"Foolish Things" is a now-and-again compilation we run based on our readers' input; once we have several of these stories assembled in one place, we get to share them with all of you. If you enjoy reading these cautionary tales of woe, proud stories of triumph, and just plain weird and fun things that happen between humans and silicon, that's great; if you have some to share so that others may enjoy them, even better. Please send them to email address.

[ You can even tell us that it happened to A Friend of Yours, and we'll believe you. ]

-- Ben

A Black Mark on Their Reputation

Karl-Heinz Herrmann

We have a nice SCSI HP ScanJet that's been sitting around for years; once in a while, we use it to scan X-ray films with a full A4 backlight. After some PC rearranging, it got moved and stopped working. The PC to which it was attached had SCSI cables that were too long (which worked fine before), but, after being shifted, the PC's SCSI disk would only run at lowest async speed. We changed about everything on the chain including the SCSI adapter in the stripped-down PC, but nothing helped. After giving up on connecting the scanner, I put in the network card again: the AGP graphics card stopped working, no screen output whatsoever. Pull the network card out: screen acts OK. Hm.

Dismissing the PC as too flaky and ancient to care about (PII 400), I decided to plug the HP scanner into my Linux laptop just to check it — and it worked perfectly. I then plugged it into a mini-tower similar to the discarded one, and it worked fine. So all is fine, right? Not so. After moving it around on the table and arranging the cables nicely, it wouldn't work. Then I realised that the scanner wasn't making the "normal noises" on power up: it would move the sledge inside a short way — and then nothing. When it was working, it would move the scan head back and forth for a few seconds, and then switch off the light. That was when we decided the scanner had a real problem after all, and sent it to a repair shop — but they sent it back saying the board was defective and they couldn't get a replacement, since it's too ancient.

There it was, sitting on the desk — a huge thing, sturdy beyond anything, and the board is "gone" just like that? By moving it a bit? I decided to open it up and look for a bad contact somewhere, maybe close to the SCSI connector. After I got the case open (tricky to find the last two screws), I saw a bit of the circuit board and the flat cables. Looks flawless to me — fine mechanics, cable guides everywhere. So, I switch it on, the head moves a little and stops with the light on — so it's still not OK. But the light is blinding me, so I put a sheet of paper on it — at which point, the head moves back and forth, goes to the 'park' position and switches off the light. (??!) It was reproducible, too. I turn around the top cover, and there is the white strip used for calibrating the scan sensors before each scan. I rub my finger over it, and it comes up black; I clean the glass plate from the inside, put it on, and — the scanner works perfectly. Damaged board indeed.... These fellows didn't even open it up.

Letting Out the Magic Smoke

Ben Okopnik

This is a story that I'm telling on myself rather than exposing someone else's foolishness; as they say, "it was in a far country, and besides, the wench is dead" — and I'm smarter (I hope!) than I was then. Many, many years ago — those of you who were working with computers then will have a good idea of when, once you read this — when I was just a PFY and still learning the computer repair trade, I was working on a relatively new machine that I had brought home from a client's place. The problem was that the motherboard manufacturer (long since out of business — and with good reason) used tiny rivets to connect the layers on the opposite sides of the board instead of properly plating everything all the way through. Well, as time went on, and thermal expansion flexed the board microscopically, these rivets came loose — and so did the connections. Randomly. Sometimes yes and sometimes no — and sometimes maybe; depending on the weather, the humidity, how recently the computer had been pounded with a fist, etc.

Expensive as it was back then — and it was very expensive indeed — I had convinced this client to replace all these boards. He agreed, through gritted teeth, but got me to promise that I would try my best to keep his old boards working as long as possible..., so I spent a lot of time soldering these rivets, using a contact-cleaning file around their edges to get at least some kind of connection, or trying to solder hair-fine wires to the traces. (This worked very rarely, since the traces were so thin that they'd burn up as soon as you touched them with a soldering iron.)

Back then, the standard bit of advice was to leave the computer plugged in but turned off while you worked on it; this provided a local ground that you were always touching and decreased the chance of blowing the chips — which were very sensitive to static back then. The board that I was working on was fairly new, and had just started showing that random behavior. As usual, I plugged it in, made sure it was turned off, and was twiddling my tiny file while quietly cursing to myself — when suddenly, the chip next to the rivet, which I just happened to touch with the file, EXPLODED. I mean, literally — blew out a chunk of its ceramic packaging so I could see the silicon underneath.

I was completely flabbergasted. I had made absolutely sure that I switched it off; I knew that the capacitors in the PC power supply were designed to discharge in a fraction of a second after shutdown... What happened? I grabbed my voltmeter and started poking around in the machine, and figured it out after about two hours of measurements:

The computer I was working on had been designed with a brand-new feature: a power switch that "put it to sleep" instead of turning it off.

(Yep... it was that long ago.)

I felt really awful, but... I did indeed follow the agreement I had with that client: that board couldn't be repaired anymore, and needed to be replaced. I forgot to mention that I was the one that contributed to its slightly earlier demise.

P.S. I never signed up for that kind of serfdom anything like that again - and I always used a grounding strap instead of a plugged-in chassis since that day. :)

Orientation Week

Rick Moen

The time having come to concoct a replacement for my server-grade 486DX2 EISA/VLB-bus Linux host, I decided one day to build a new machine from best-of-breed parts (well, within limits set by my being a cheapskate).

I happened to have an excellent tower case and PC Power & Cooling power supply handy, a pair of Adaptec PCI SCSI host adapters, a SoundBlaster Pro, Matrox G200 video, a couple of large, fast IBM SCSI hard drives, a Plextor SCSI CDR, and a pair of 3Com 3C509B 10 megabit ethernet cards. (Hey, all it had to do was potentially max out a 1.54 megabit T-1 line, OK?) That left the matter of motherboard and parts attendant thereto.

After some checking, I bought a nice little FIC PA-2007 Socket-7-type motherboard. Then, from my favourite vendors (SA Technology), I acquired 128 MB of Crucial Technology SDRAM capable of running at CAS2 operation at 100MHz, an AMD K6/233 CPU, and a big-kahuna heat sink with a fan on top that uses ball bearings instead of the standard, noisy, failure-prone sleeve bearings. The ensemble was designed to be highly reliable, and at the same time hackable if I ever decided to join the then-pervasive overclocking madness.

Anyhow, I banged everything together. It seemed to work fine. The big tower case, low-heat-output CPU, and major cooling capacity meant that the thing ran very cool and reasonably quiet, even with the two very fast IBM 10,000 RPM SCSI hard drives.

As was my custom, I started compiling a kernel after a bit. That compile errored out with a SIG11. Hmm. Tried again. Errored out even faster, and at a different point in the compile. Odd. Do I have bad RAM?

Shut down and re-seated the RAM. Went downstairs to the CoffeeNet (a recently reborn 1990s Linux-based Internet cafe I'd helped build) for a caffeine infusion. Came back, powered up, ran the compile. No problems. Ran the compile again: SIG11. Ran it again: SIG11 at a different place. Ran it with only half the RAM: Same symptoms. Ran it with the other half: Same. Didn't seem to be a RAM problem(?).

Slept on the problem. Woke up, fired up the machine, ran a compile: No problem. Ran it again: No problem. Ran it a third time: SIG11. Ran it again: Same error, different spot.

Pondered the problem for a bit: It seemed as if the error was kicking in only after the system reached heat equilibrium, but not during the initial 30 minutes of operation when the system was still stone-cold. But that didn't make much sense: I opened up the case and re-verified that the system really was an engineer's dream of conservative design, and that even the 10,000 RPM drives were running cool.

I drove down from San Francisco to the Palo Alto Fry's and bought both the heat-conductive pads you can sandwich between CPUs and their heat sinks and the thermal paste you can use instead. I was going to make double-sure that I had good contact, there, before taking the CPU back to SA Technology and looking like an idiot if it turned out to be perfectly OK.

I took the heat sink off the CPU, cleaned both off, put paste on them and pressed them together. For some reason, perhaps on account of some bell ringing in my unconscious mind, a few minutes later I pulled them apart again to look at the two pieces more closely.

And swore.

The Socket-7 motherboard socket for the K6 has a square outline, and the pins have a square pattern, with one corner of the CPU's pin-layout being different so you won't destroy it by putting it in the wrong way. When you put the CPU into the socket, if you aren't looking too closely (cue ominous music), you think: square socket, square CPU with a keyed feature to keep you from screwing up, square heatsink/fan assembly. 1, 2, 3 — done. Foolproof. (Well, no.)

The top surface of the CPU turned out to have a heatsink-contact surface that extended only across maybe 70% of the lateral distance, and the other 30% was sunk down lower.

|            |      |
|            |      |
|            |      |
|            |      |
|  contact   |      |
|  surface   |      | 
|            |      | 
|            |      |

I'd accidentally rotated the heatsink 180 degrees, so that its contact surface was staggered over to the right-hand-side:

|      |            |
|      |            |
|      |            |
|      |            |
|      |   contact  |
|      |   surface  |
|      |            |
|      |            |

Only a little strip in the middle was actually touching, so all the other thermal paste was still protruding up into the air, un-squashed. Most of the CPU top surface was getting zero help with cooling, and instead was radiating out into a nice warm, insulating air pocket under the heatsink.

If this had been an Athlon or P4 Coppermine, the CPU probably would have committed seppuku, but the K6 was completely undamaged and has been happily cranking away ever since.

But that experience confirmed me in my prejudice that a cool CPU (like a cool system generally) is much, much, much to be preferred over one that runs hot and needs heroic measures like huge amounts of forced air flow to stave off disaster. Those other ones may be faster — but usually (for most machine roles) don't even manifest that speed in ways you especially care about.

Talkback: Discuss this article with The Answer Gang


Ben is the Editor-in-Chief for Linux Gazette and a member of The Answer Gang.

Ben was born in Moscow, Russia in 1962. He became interested in electricity at the tender age of six, promptly demonstrated it by sticking a fork into a socket and starting a fire, and has been falling down technological mineshafts ever since. He has been working with computers since the Elder Days, when they had to be built by soldering parts onto printed circuit boards and programs had to fit into 4k of memory. He would gladly pay good money to any psychologist who can cure him of the recurrent nightmares.

His subsequent experiences include creating software in nearly a dozen languages, network and database maintenance during the approach of a hurricane, and writing articles for publications ranging from sailing magazines to technological journals. After a seven-year Atlantic/Caribbean cruise under sail and passages up and down the East coast of the US, he is currently anchored in St. Augustine, Florida. He works as a technical instructor for Sun Microsystems and a private Open Source consultant/Web developer. His current set of hobbies includes flying, yoga, martial arts, motorcycles, writing, and Roman history; his Palm Pilot is crammed full of alarms, many of which contain exclamation points.

He has been working with Linux since 1997, and credits it with his complete loss of interest in waging nuclear warfare on parts of the Pacific Northwest.

Copyright © 2006, Ben Okopnik. Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 128 of Linux Gazette, July 2006