Talkback

Talkback

Talkback:124/smith.html (10)

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

Amber Sanford (amber at modernspaces.com)
Thu Aug 17 11:41:25 PDT 2006

[Ben] - You bet: switch to Linux. :)

Amber, I'm not the world's top expert when it comes to Windows, but I've got quite a lot of experience and expertise with it under my belt. Speaking from that perspective and to the best of my knowledge, there are a few minor things that you can do in this direction - i.e., a pair of video cards and even multiple keyboards attached to a single CPU - but all of this hardware is still tied to one session, i.e. one person using all of it. There are some uses for this type of configuration, but it's a completely different kettle of fish.

As far as creating a robust, serious piece of software that will do this under Windows, or even a project toward that end, it's simply not going to happen: Microsoft would need to open-source their OS, and that's not in the cards. This is one of the many reasons that so many people and so many companies are switching to Linux: in the world of Open Source, if you have this kind of a requirement, you can either find someone somewhere who has already done it (and perhaps fund them to tweak it to your exact specs) - or you can do it yourself (perhaps by hiring a little programming muscle if the capability does not exist in-house.) It's a different - and we believe, better - approach to computing.

Best of luck in your search, and we'll be happy to hear from you if you decide to make the switch at some point in the future.

Talkback:124/smith.html (11)

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

José Antonio (jap1968 at yahoo.es)
Tue Sep 12 15:39:45 PDT 2006

[Ben] - That's a good article, José - thank you. I'm going to CC the author of the piece you're responding to; perhaps you two can discuss ways in which you can help each other, or share the knowledge you've gained in the process of doing these projects.

Talkback:124/smith.html (12)

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

tkalenko_ma (tkalenko_ma at sibadi.org)
Sun Sep 17 19:41:09 PDT 2006

[Ben] - (Comment for Mike Orr: Oh boy! This is your big chance to practice some of your "Russian humor" on me! Go ahead, I'm ready. :)

> I have read your article about 6-user system.
> How possible configuring similar system on two doubleheaded PCI-E cards
> (SLI motherboards) + one doubleheaded PCI card?

You have very interesting timing, Maxim; one of our readers, José Antonio, just wrote in describing a system that's built around dual-video cards. I suspect that what you're asking is possible; take a look at the link below and experiment. We would appreciate being kept apprised of your progress - judging from the amount of mail we've received about Bob's article, many people seem to be interested in this and similar issues.

http://netpatia.blogspot.com/2006/09/multiseat-computer-with-ubuntu.html

Good luck!

[BobS] - I'm sorry that I can not give you any useful advice in setting up your dual head PCI-E card. Ben is right, the article he references may help.

Talkback:126/howell.html (3)

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

S.K.Goel (skgoel at omlogistics.co.in)
Wed Sep 20 21:32:31 PDT 2006

[Neil] - My advice is to install it and get stuck in.

There are forums at http://www.opencobol.org/, which will no doubt be better placed to help you with the transition than the members of this list.

Talkback:128/adam.html

[ In reference to How Fonts Interact with the X Server and X Clients in LG#128 ]

Thomas Adam (thomas.adam22 at gmail.com)
Tue Sep 12 05:16:54 PDT 2006

http://lwn.net/Articles/189901/

Posted Jul 3, 2006 17:41 UTC (Mon) by subscriber otaylor

Probably not news to most LWN readers, but the font article should be
ignored: it is describing technologies that are no longer in common
use. Goodbye and good riddance to the XLFD (X Logical Font
Description). Contemporary applications, toolkits, and desktops use the
fontconfig library instead.

Talkback:128/ramanathan.html

[ In reference to Subversion: Installation, Configuration — Tips and Tricks in LG#128 ]

Nathaniel Ye (Nye at airliteplastics.com)
Tue Jul 18 09:22:40 PDT 2006

[root at localhost test]# ldd /usr/lib/httpd/modules/mod_dav_svn.so | grep apr
        libaprutil-0.so.0 => /usr/lib/libaprutil-0.so.0 (0x00cb2000)
        libapr-0.so.0 => /usr/lib/libapr-0.so.0 (0x00man 6a5000)
[root at localhost test]# ldd /usr/sbin/httpd | grep apr
        libaprutil-0.so.0 => /usr/lib/libaprutil-0.so.0 (0x002ba000)
        libapr-0.so.0 => /usr/lib/libapr-0.so.0 (0x005ca000)

[root at localhost lib]# ls -al libapr*
lrwxrwxrwx  1 root root     17 Jul 17 10:22 libapr-0.so.0 -> libapr-0.so.0.9.4
-rwxr-xr-x  1 root root 139868 May 17  2005 libapr-0.so.0.9.4
lrwxrwxrwx  1 root root     21 Jul 17 10:23 libaprutil-0.so.0 -> libaprutil-0.so.0.9.4
-rwxr-xr-x  1 root root  83260 Jun 16  2005 libaprutil-0.so.0.9.4

[Ramanathan] - You should not change the soft links to newer versions.

You can try to recompile svn and use the --with-apr switch in the configure script to point to the apache version of the apr. apxs can also be a problem.

Talkback:129/okopnik1.html

[ In reference to Low-Fat Linux - Now with Less Cruft! in LG#129 ]

Mark Baldridge (mbaldrid at us.ibm.com)
Sat Sep 2 18:09:48 PDT 2006

[Ben] - Sure - in '/var/log', just like many other *nixen. Those aren't much of a problem, though: most programs that create logs also create an entry in '/etc/cron.{d,daily,monthly,weekly}' which rotates those logs at that specified interval.

There's also the fact that this is one of the classic reasons for having multiple partitions: even if '/var' does get filled up, '/' isn't affected.

[Faber] - Or, on Red Hat derived boxen, the application puts an entry in /etc/logrotate.d and let's the logrotate daemon do the heavy lifting.

BTW, I've had the joy recently to work on a Debian box. A non-standard one at that. Boy, you guys do things weirdly!

[Ben] - Why, Faber... you've never struck me as a religious type before. Wanna argue about Emacs vs. Vim next? :)

Pot. Kettle. #000000. :)))

[Faber] - Why bother? Everyone knows that pico rules!

:-)

[Kapil] - Actually "nano" is better than "pico" (smaller is better).

(Imagine devilish grin; I have forgotten the emoticon)

[Rick] - Wimp.

$ ls -l /etc/alternatives/editor
lrwxrwxrwx 1 root root 12 2006-03-10 14:28 /etc/alternatives/editor -> /bin/cat

[Ben] - Wuss.

http://ars.userfriendly.org/cartoons/?id=19990508&mode=classic

[Faber] - I thought this was appropriate (wait for the nag screen to go away)

http://ars.userfriendly.org/cartoons/?id=20060904

[Ben] - [laugh] Great minds think alike, of course. Two great cartoons...

> BTW, I've had the joy recently to work on a Debian box. A non-standard
> one at that.  Boy, you guys do things weirdly!

On a Debian system /etc/logrotate.d also works the same way so I am mystified by your complaint---if it was one!

[Rick] - Maybe some Debian-oriented BOfH is pulling a prank on you?

$ ls -l etc/logrotate.d
total 13
-rw-r--r-- 1 root root  366 2005-01-19 19:32 apache
-rw-r--r-- 1 root root  240 2006-01-16 02:15 apache2
-rw-r--r-- 1 root root   79 2004-09-28 11:44 aptitude
-rw-r--r-- 1 root root  384 2004-12-03 14:25 base-config
-rw-r--r-- 1 root root  111 2005-09-26 00:04 dpkg
-rw-r--r-- 1 root root  170 2005-01-05 02:07 exim4-base
-rw-r--r-- 1 root root 1272 2005-01-14 01:23 mailman
-rw-r--r-- 1 root root 1072 2005-09-29 15:19 mysql-server
-rw-r--r-- 1 root root 1020 2005-01-18 14:44 mysql-server.dpkg-old
-rw-r--r-- 1 root root  128 2004-11-08 14:18 super
-rw-r--r-- 1 root root  134 2004-07-11 21:08 vsftpd
$

[Faber] - I'm leaning towards "incompetent"; there are kernels, initrds and log files in /, there's a directory called /images that holds CSS files, etc.

[Ben] - It may have started life as a Debian system, but that's not the right description for it any more. I think that somewhat stronger terms are much more applicable. :) Debian follows the FHS these days (http://qa.debian.org/fhs.html), and has for a while now.

Talkback:130/tag.html

[ In reference to The Monthly Troubleshooter: Installing a Printer in LG#130 ]

Steve Brown (steve.stevebrown at gmail.com)
Sun Sep 3 14:07:10 PDT 2006

[Ben] - Hi, Steve -

Great story - much in the vein of our erstwhile "Foolish Things We Do With Our Computers" column, although in this case, it's "Foolish Things Other People Do With Their Computers." As to the guy that you're talking about - I've met him! Or, well, people who are exactly like him... lots of them. Tech support will do that to^Wfor you. :)

Thanks for writing!

Talkback:130/neville.html

[ In reference to DNS techniques in LG#130 ]

Blizbor (tb670725 at ima.pl)
Thu Sep 7 06:50:08 PDT 2006

[Thomas] - On 07/09/06, Blizbor <tb670725 at ima.pl> wrote:

> I wish you Linux Gazette never again put such poor quality material on
> your pages.

You're absolutely right -- why, when I look back at the hundreds of issues, I can see several mistakes, all of which have large arrows which point to the nasty people of LG. What awful creatures we are at LG. Tut.

> Fix this asap - only those doing nothing arent doing mistakes.

No -- to do so is pointless since the mirrors of LG would have already have taken the tarball themselves. To change it at lg.net (which may or may not happen -- it's not my say in such matters) is the best we can do.

[Ben] - [ Forwarded to author. ]

On Thu, Sep 07, 2006 at 03:50:08PM +0200, Blizbor wrote:

> There are at least one principal mistake done - in bind you can 
> control caching nameserver. Telling the trueth - problem is in
> complexity of used tools rather than one is incapable for something.
> I strongly suggest reedition of this article in context: "how it could
> be done using bind, and how uisng djb" leaving choice of solution to users.

You're welcome to write such an article.

> To the author - in the bind manual is a chapter about "view" keyword, I
> appreciate your work in writing this article - well done about djb, but
> you missing trueth about bind.  In actual form article is unacceptable
> and should be removed.

Thank you for your opinion. I take it you're volunteering as a technical editor to the Linux Gazette? If you do, and then succeed in establishing some credentials for your technical knowledge, your opinion will be considered along with others here.

> I wish you Linux Gazette never again put such poor quality material on
> your pages.

Then I suggest you get cracking on that application. I'll be waiting for it with baited breath.

> Fix this asap - only those doing nothing arent doing mistakes.

The answer, then, is for you to stop doing nothing and contribute your time and effort to helping LG and the Linux community.

> (This is my personal opinion.)

Really? In that case, here's mine: if you haven't contributed, don't be so quick with the harsh criticism.

[Rick] - Quoting Blizbor (tb670725 at ima.pl):

> There are at least one principal mistake done - in bind you can
> control caching nameserver.  Telling the trueth - problem is in
> complexity of used tools rather than one is incapable for something.

Greetings, Blizbor. I can perhaps comment as the editor who did the technical edit of Ed Neville's article (whom I am cc'ing). You might recognise my name from the several footnotes I added.

On the matter of controlling BIND9's caching nameserver functionality, please note that Mr Neville correctly and commendably qualified his statement by saying that open access to caching is a problem of BIND9's _default installation_. That is absolutely correct as stated, and a very valuable point that should be heeded by all BIND9 users.

> I strongly suggest reedition of this article in context: "how it could
> be done using bind, and how uisng djb" leaving choice of solution to
> users.  To the author - in the bind manual is a chapter about "view"
> keyword, I appreciate your work in writing this article - well done
> about djb, but you missing trueth about bind.

That would have been a very different article from the article Mr Neville chose to write; _Linux Gazette_ has no wish to dictate the scope of articles to authors. We merely ask that they be clear and accurate, so as to increase the wealth of understanding among our readers, and are delighted to assist authors in hitting that target.

Along those lines, you seem to have missed the fact that BIND9's 'view' keyword actually is described within Mr Neville's article, courtesy of the additional material I added in footnote number 1, referring any interested readers to Rob Thomas's 'Secure BIND Template' for good examples.

> In actual form article is unacceptable and should be removed.  I wish
> you Linux Gazette never again put such poor quality material on your
> pages.  Fix this asap - only those doing nothing arent doing mistakes.
> 
> (This is my personal opinion.)

I say this as a sysadmin familiar both with DJBware and with its open-source alternatives, who chooses to use BIND9 for nameservice on various servers, and who has published in LG a primer on simple DNS setup modes in BIND9 (http://linuxgazette.net/121/moen.html): Bosh. Bollocks. Mr Neville's piece was technically adept and well written.

(It also may be of interest that I'm a major devil figure for many of the less civil of DJB's coterie, because of critiques I have written online. E.g., Prof. Bernstein's Web pages call me names on account of those critiques.)

[Ed] - Rick Moen <rick at linuxmafia.com> wrote:

> Quoting Blizbor (tb670725 at ima.pl):
> 
> > There are at least one principal mistake done - in bind you can
> > control caching nameserver.  Telling the trueth - problem is in
> > complexity of used tools rather than one is incapable for something.

Can you explain any other errors you feel need resolving, if there are technical mistakes it might be possible to alter it, but I think the only point you want emphasised is that small mention of BIND.

> please note that Mr Neville correctly and commendably qualified his
> statement by saying that open access to caching is a problem of
> BIND9's  _default installation_.  That is absolutely correct as
> stated, and a very valuable point that should be heeded by all BIND9
> users.

I knew there would be some flack if I did not! I've had this same old discussion all over the place. The article is in no way about "my NS is better than your NS", what I wrote is my experience of running a large ISP's NS, it's what works well for us and our customers. The introduction was to explain why I wrote it, if nothing else, the various components helps the reader to understand a little about lookups in the process.

> I say this as a sysadmin familiar both with DJBware and with its
> open-source alternatives, who chooses to use BIND9 for nameservice on
> various servers, and who has published in LG a primer on simple
> DNS setup modes in BIND9 (http://linuxgazette.net/121/moen.html):  
> Bosh.  Bollocks.  Mr Neville's piece was technically adept and well
> written.
> 
> (It also may be of interest that I'm a major devil figure for many 
> of the less civil of DJB's coterie, because of critiques I have
> written online.  E.g., Prof.  Bernstein's Web pages call me names on
> account of those critiques.)

Oh! You're that /Rick Moen/, I would not have know unless you had pointed it out, quite a mean thing DJB did there putting your mail address and name on the FAQ.

[Rick] - Quoting ed (ed at s5h.net):

> I knew there would be some flack if I did not! I've had this same old
> discussion all over the place. The article is in no way about "my NS is
> better than your NS", what I wrote is my experience of running a large
> ISP's NS, it's what works well for us and our customers. The
> introduction was to explain why I wrote it, if nothing else, the various
> components helps the reader to understand a little about lookups in the
> process.

If I may say so, I've learned a great deal from study of djbdns and from reading technical pieces written by knowledgeable members of the DJBware community. In particular, I'll treat Jonathan deBoyne Pollard and Russ Nelson to tall quaffs from their favourite beverages any day of the week, out of gratitude. You follow in their footsteps, and I'm glad to have "met" you.

> Oh! You're that /Rick Moen/, I would not have know unless you had
> pointed it out, quite a mean thing DJB did there putting your mail
> address and name on the FAQ.

Well, it gives me a rare distinction, actually: I can ask people who else they know who's mentioned by name in a major software licence? ;-> The mean-spritedness doesn't bother me, in fact, but Dan's non-sequitur evasion of my substantive critique did. (Please pardon this URL, as I was a bit annoyed at the time:) http://linuxmafia.com/~rick/faq/just-another-djb-groupie.html

(Sometimes, I get asked on IRC 'Are you really Rick Moen?', to which my traditional answer is 'No, just someone else of the same name.')

> Quoting Blizbor (tb670725 at ima.pl):
>
>   
>> There are at least one principal mistake done - in bind you can
>> control caching nameserver.  Telling the trueth - problem is in
>> complexity of used tools rather than one is incapable for something.
>>     
>
> Greetings, Blizbor.  I can perhaps comment as the editor who did the
> technical edit of Ed Neville's article (whom I am cc'ing).  You might
> recognise my name from the several footnotes I added.
>
> On the matter of controlling BIND9's caching nameserver functionality, 
> please note that Mr Neville correctly and commendably qualified his
> statement by saying that open access to caching is a problem of BIND9's 
> _default installation_.  That is absolutely correct as stated, and a
> very valuable point that should be heeded by all BIND9 users.

[Rick] - Quoting Blizbor (tb670725 at ima.pl):

> I'm a bit not precise in saying what I mean. Sometimes what I want to
> say are a bit unkind.

It's OK. We try to work through to the substance, which is the important thing.

> The point of my mail is: nobody on the world is using _default
> configuration_.

I wish that were true, but for example a presentation by Dan Kaminski at the 2005 LISA Conference in San Diego revealed the interim results of his project to study the world's Internet-reachable DNS servers -- including the fact that a frighteningly large percentage are vulnerable to cache poisoning: I think the figure was well over 25% (would have to check).

You can also spot-check domains you know: I think you'll find the problem to be widespread. For example, I just tried http://www.dnsreport.com/tools/dnsreport.ch?domain=cocacola.com . Notice that both of the Coca Cola Company's are open to public recursive queries. Nameserver software versions were not available in that case, but Kaminsky's results suggest that BIND8 versions for Unixes and Windows still predominate.

> Actually I think that default configuration of any network demons
> should be crippled to the extent they are do start and do extremely
> limited functionality on the loopback interface.

I tend to agree with you, but that is not Mr Neville's responsibility, especially since his piece wasn't about BIND9 in the first place. To the extent he referred to that software, his point was correct, well stated, and properly qualified.

> I found referring to default configuration as primary source of
> principal mistake of the article.

We will have to agree to disagree, since I see no mistake. To the contrary, Mr Neville was commendably careful -- and, incidentally, much more helpful to BIND9 users than are many who actually write on that subject.

> It can be done and should be said "it can be, but we will keep focus
> on how to do that using djbdns because ...".

I've read advocacy pieces, and this wasn't one.

> Wait ? I read that article and why exactly it's worth to use DJB ... I
> still don't know.

I'd say the piece was more how to effectively use djbdns. It should be read in that spirit.