Debian Traffic #14 For 14 Dec 2000

Editor: Zack Brown

By Peter EckersleyPrashanth Mundkur  and  Zack Brown

Debian Home Page (http://www.debian.org) | Weekly News (http://www.debian.org/News/weekly/) | Social Contract (http://www.debian.org/social_contract) | Constitution (http://www.debian.org/devel/constitution) | Policy Manual (http://www.debian.org/doc/debian-policy/)
Developer's Reference (http://www.debian.org/doc/packaging-manuals/developers-reference/) | Documentation Project (http://www.debian.org/doc/ddp) | debian-devel Archives (http://lists.debian.org/#debian-devel)

Table Of Contents

Introduction

Want to help write KC Debian? See the KC Authorship page (../author.html) the KC Debian homepage (index.html) , and the Thread Summary FAQ (../summaryfaq.html) . Send any questions to the KCDevel mailing list. (mailto:kcdevel@zork.net)

Mailing List Stats For This Week

We looked at 1416 posts in 5048K.

There were 360 different contributors. 202 posted more than once. 0 posted last week too.

The top posters of the week were:

1. Converging the Debian build

22 Nov 2000 - 26 Nov 2000 (9 posts) Archive Link: "autobuilding and embedded timestamps"

Summary By Prashanth Mundkur

People: Edmund Grimley EvansDaniel JacobowitzWichert Akkerman

Edmund Grimley Evans started a short thread when he asked, " Do Debian's autobuilders compile repeatedly until the system converges? If so, how do they detect convergence? " He had " been playing around with a small system [using 20 source packages] that can compile itself. I like to make the system compile itself until it converges, that is, the newly compiled system is identical to the installed system." He added that embedded timestamps in the .a archives, lilo and the kernel made it hard to check this sort of convergence, by causing the same source to give rise to a different binary each time it is compiled.

Daniel Jacobowitz answered the question with: " Heck no. Nothing gets built more than once. It causes all sorts of problems, but convergence is essentially impossible - we're building ALL of the debian archive, remember? We can't pick and choose 20 packages; we have to deal with 4000 instead." When Edmund said that the build time would be linear in the number of packages, giving rise to a mere 4000 (!) extra builds for each test cycle, Daniel Jacobowitz, shocked at the suggestion, pointed out: " A complete build of Debian, on a top-end machine, would probably take three or four complete days. [...] That's not really feasible, and every time you needed to redo a cycle...."

There was some further discussion about the meaning of embedded timestamps between Edmund and Wichert Akkerman, and the thread died.

2. Debian on Sony Vaio laptops

25 Nov 2000 - 29 Nov 2000 (16 posts) Archive Link: "sony vaio laptops"

Summary By Prashanth Mundkur

People: Jordan EvattAdam McKennaBen GertzfieldRobert van der MeulenDaniel JacobowitzBernd EckenfelsMartin Schulze

Jordan Evatt, looking to replace his disintegrating laptop, was planning to go with a Sony Vaio, and said: " I obviously want to make sure the Vaios have good video graphics capabilities and excellent linux support for all that is included with them." and asked for success stories installing/using Debian on them.

Although Adam McKenna was delirious about potato and woody on his F480: " Everything works, including sound and X4" , some problems quickly emerged from the woodwork. Ben Gertzfield described the problems he had with his N505VE, which required a kernel parameter on CD-install, and " There are MAJOR problems with the N505 series and APM -- after a suspend, sound and PCMCIA stop working." Robert van der Meulen responded with his favorable experiences running 2.4.0-test8, with working PCMCIA after suspend. For sound, Robert said: "Sound didn't [work] at first, but i suspend trough a script, and unload my sound drivers first (and reload them on resume). That has been working for ages now." Ben then updated to kernel 2.4.0-test11, but again experienced the earlier broken PCMCIA behaviour after suspend, and asked for the build .config file in private email. There was no further mail from Ben on whether that worked out.

Daniel Jacobowitz jumped in with: " my evil workaround to this is to always suspend to disk (hibernate) instead of suspending to RAM. That'll not break (or even fix, if you suspended to RAM earlier) PCMCIA."

Elsewhere, Ishikawa Mutsumi provided the kernel boot parameter to access the CD-ROM drive during install: ide2=0x180,0x386.

Martin Schulze mentioned the physical fragility of these laptops and their breaking with usage. Bernd Eckenfels chimed in with:

The same is true for my Wife's F304. The Display-mechanism is just a thin metal blade which will break off the computercase if it gets stuck for some reason while closing the display.

And also the repair for the keyboard which started to lose single keys took month. This is everything espeiclly annying since the notebook was never in heavy use and always well carried in a bag :(

3. Debian Job Site

29 Nov 2000 - 30 Nov 2000 (7 posts) Archive Link: "job offers"

Summary By Zack Brown

People: Michael MeskesVince Mulhollon

Michael Meskes asked:

Do we have anything like a job listing on our website? Or is there any other website listing Debian related jobs? If not how about creating one?

Actually I came to this idea because we are urgently looking for some guys especially for part time jobs (students etc.) on and with Debian. So the logical choice to look at would be www.debian.org or www.debian.com or somesuch. But it appears nothing is available.

I would be interested in creating one but of course that adds to the work we have waiting here, so I need more guys, so I need the web page more urgently, ...

Anyway, what do you think? Would it be a good idea to add a webpage with job offerings?

Peter Novodvorsky, David Engel and Andreas Mueller loved the idea, and Andreas added that he'd already made such a thing for his local Debian group. He gave a URL for a page in Turkish (http://tr.debian.net:88/job.php3) ; and Michael asked if his code had been packaged yet (and if it was available in other languages. Elsewhere, Vince Mulhollon replied to Michael's initial post, with:

I forwarded the idea of a "debian jobs list" to debianplanet.

I think the only problem or issue to consider would be to make it clear that "debian.org does not endorse any particular employer or employee" etc.

Michael suggested using debian.com for this sort of thing (currently it resolved back to debian.org.

4. Status of Open Office packaging

30 Nov 2000 - 2 Dec 2000 (8 posts) Archive Link: "StarOffice / Open Office progress"

Summary By Prashanth Mundkur

People: Joseph CarterEray OzkuralAdrian Bunk

Paul Hedderly wrote in to ask about progress in packaging of OpenOffice, which was released on CVS back in October. Joseph Carter, who's handling the packaging singlehandedly, replied:

StarOffice is a BIG honking package and I am very likely to package it for Progeny as soon as the benefit is worth the effort (ie, it can be cleanly built and actually runs) and will be happy to upload the package to Debian. In fact, if I'm going to go through the effort you damned well better believe I intend to upload it to Debian.. =)

When things are a little less stressful, I'll have another look and see what I can do with it.

When Eray Ozkural asked whether somebody else should take a look at it in the meantime, Joseph said they were welcome to, but added "given the state it was in when Eric and others looked at it, it's not ready to be useful on its own yet. Of course that was a couple of months ago and a lot may have changed since."

On hearing this, Adrian Bunk, who had earlier offered to package it, offered to have a look in a week.

5. Patches to the Debian kernel

30 Nov 2000 - 4 Dec 2000 (12 posts) Archive Link: "kernel with supermount"

Summary By Peter Eckersley

People: Massimo Dal ZottoJosh HuberSean 'Shaleh' PerryJoseph CarterHerbert Xu

Massimo Dal Zotto requested that the supermount patch be included in the next Debian kernel, saying "I've been using a supermount kernel with potato from months and it works very well." Josh Huber responded, saying "Along similar lines, would it be reasonable to include the rawio patch for the next release?"

Sean 'Saleh' Perry replied, saying "I expect this [rawio] to appear in upstream kernels fairly soon. The less work Debian has to do to test kernels the better. By using a fairly stock kernel, we are not as prone to weird Debian only bugs. Also, users tend to grab kernels from the upstream just as often as they use the one shipped by Debian. If functionality suddenly disappears when the user thought they were upgrading, it would be bad."

Several developers supported the "as few patches as possible" philosophy, while Massimo defended his position and Joseph Carter commented, "I must agree that more or less every major distribution except us supplies a rather heavily patched kernel. In truth, the reason not to patch our kernel is more in that maintaining our patches is a PITA with the current source packages and that dbs is no substitute for a modern source package format. =p"

Herbert Xu, the Debian kernel maintainer, replied "FWIW, my policy is to minimise the amount of new features, while including any bug fixes which are necessary. Oh, and CVS (which I use for all my packages) has no problems dealing with a large amount of patches."

It appears unlikely that a flurry of patches will adorn the Debian kernel in the near future.

(ed. [] It's a pity there's no easy way of adapting "popularity contest" so that widely applied patches can be identified and included :) )

6. Planning A Woody Freeze

3 Dec 2000 - 4 Dec 2000 (36 posts) Archive Link: "Woody Progress"

Summary By Zack Brown

People: Julian GilbeyRoland BauerschmidtDavid Z. MazeAnthony TownsBas ZoetekouwJohn GaltRichard Taylor

Richard Taylor asked when Woody would become stable, and suggested that sooner would be better than later. Several folks replied to suggest shorter release schedules in general, and Julian Gilbey also added, "the new "testing" distribution should be working within the next couple of months. It will be like unstable, except that it will lag something like two weeks behind, so that badly broken packages don't make it into testing." Elsewhere, Roland Bauerschmidt remarked, "IIRC Anthony plans on freezing in January and having a quick freeze. Is this correct?" David Z. Maze replied, "It's worth noting that slink and potato both had "quick" freezes that went on for 3+ months; if this is in fact the plan, I'd be interested in hearing how to make it "quick"." Anthony Towns clarified his situation, "The date I made up was mid February, ie, six months after the potato release. That's only indicative, not set in stone." He added:

http://auric.debian.org/~ajt/ contains an automatically generated selection of packages from amongst potato and woody that is roughly as release-worthy as potato, and contains as many new packages from woody as possible.

Working out how to use that to make a freeze "quick" is left as an exercise to the reader.

Regarding the release schedule, Bas Zoetekouw replied, "Wouldn't it be a good idea to set a definate freeze date now? That way, people can plan on what that want to do with their packages before the freeze." Thereabouts, John Galt replied historically, "There was no Debian 1.0--just before Debian went 1.0, a vendor jumped the gun and had CDs pressed before the release date. That's why we have codenamed releases like bo, hamm, slink, potato, etc. There will hopefully never be another definite freeze date, only release goals." Rando Christensen replied that he didn't think that principle should apply to freeze dates, which could only help developers, in his opinion. John agreed, as did Bas, who added, "The Freeze date is something internal, while is release date is not. The freeze should indeed last as long as it takes (until we are satisfied with the result). The time before the freeze, however, does not need to last: on the contrary: a definate freeze date would allow developers to do a bit of planning." The thread ended there.

7. Freenet For Debian

3 Dec 2000 - 5 Dec 2000 (3 posts) Archive Link: "ITP: freenet -- a peer-to-peer networking daemon with anonymity"

Summary By Zack Brown

People: Evan ProdromouStephane Bortzmeyer

Evan Prodromou announced his intention to package Freenet (http://www.freenetproject.org/) , a tool for anonymously distributing arbitrary files across the internet. From the description:

Freenet is a peer-to-peer network designed to allow the distribution of information over the Internet in an efficient manner, without fear of censorship. Freenet is completely decentralized, meaning that there is no person, computer, or organisation in control of Freenet or essential to its operation. This means that Freenet cannot be attacked like centralized peer-to-peer systems such as Napster. Freenet also employs intelligent routing and caching meaning that it learns to route requests more efficiently, automatically mirrors popular data, makes network flooding almost impossible, and moves data to where it is in greatest demand. All of this makes it much more efficient and scalable than systems such as Gnutella.

But Evan added, "Freenet requires a JDK 1.1-level Java virtual machine. The Debian kaffe package doesn't work, since it doesn't support big numbers." Stephane Bortzmeyer replied, "This is true for potato but the CVS upstream kaffe has fixed this problem at least several months ago and I believe it is in the woody package (I use Freenet with kaffe)." Evan replied:

Well, as far as I know, neither your bug report...

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=66064

nor mine...

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=78544

have been addressed. There are a couple more in there. Mine is for version 1.06, the current woody package.

Kaffe itself supports BigInteger, but it's a configure-time option, and apparently the configuration detection doesn't work so well with libgmp3. So you can use Freenet with a home-built Kaffe, but not with the Debian package. The changelog for the kaffe package specifies that gmp dependency was -removed- recently.

End of thread.

8. Debian Auto-Install

5 Dec 2000 - 6 Dec 2000 (7 posts) Archive Link: "boot-floppies -> unattended install"

Summary By Zack Brown

People: Jonathan D. ProulxGlenn McGrathEric Gillespie, Jr.Thomas Lange

Jonathan D. Proulx said, "rumor has it someone is working on an unattended install for Debian similar to Solaris JumpStart ar RedHat KickStart." He asked about the status of this, and several folks replied. Glenn McGrath said:

There has been talk of implementing this sort of thing in the new installer that is being developed, current work on the new installer has so far been concentrated on basic functionality.

If you look in debian cvs under debian-installer there are some docs and code, discussion about it goes on in the debian-boot mailing list.

The installer team can always do with more helpers.

Eric Gillespie, Jr. also replied to Jonathan:

I don't know if this is the rumor you heard or not, but we have an auto-installer here at Progeny. There is no reason it can't exist along-side the Debian installer.

With the auto-installer, you configure one machine, and use a GUI tool (or edit the files yourself, they are human-readable) to set a few things, such as network configuration (which can use DHCP or a table of IP addresses). Then, you build a boot disk and put it in a new computer, turn it on, and walk away. When you come back, getty or XDM or whatever is waiting for you.

Also, our GUI interactive installer is built on top of that, and there's no reason that can't coexist with the new Debian installer either.

The source to all this (except the GUI auto-installer configurator, which isn't ready yet) will be available on Monday with Progeny Debian Beta 2.

If you have any more questions, you may want to talk to jlicquia@debian.org, who is the clever guy responsible for the auto-intaller.

Meanwhile, Thomas Lange also replied to Jonathan's initial question, saying, "That's me. FAI is a fully automatic installation. It's not an official Debian package yet, but maybe till end of this year or in january. It's working quiet fine, so take a look at http://www.informatik.uni-koeln.de/fai/. The next version 1.4.1 will be released this week."

 

 

 

 

 

 

Sharon And Joy
 

Kernel Traffic is grateful to be developed on a computer donated by Professor Greg Benson and Professor Allan Cruse in the Department of Computer Science at the University of San Francisco. This is the same department that invented FlashMob Computing. Kernel Traffic is hosted by the generous folks at kernel.org. All pages on this site are copyright their original authors, and distributed under the terms of the GNU General Public License version 2.0.