Log from the Debian Installer team meeting of Jan. 28th 2006, 17:00UTC ---------------------------------------------------------------------- People who speak below: bubulle : Christian Perrier chicky : Miroslav Kure fjp : Frans Pop joeyh : Joey Hess Kamion : Colin Watson luk : Luk Claes makx : Maximilian Attems _max : Max Vozeler Mithrandir: Tollef Fog Heen netman1 : ? VivekVC : Vivek Varghese Cherian waldi : Bastian Blank zinosat : Davide Viti Hours are European Time (UTC +0100): 17:55 < bubulle> Meeting in 5 minutes. People attending, now you can say hi to confirm you're here... 17:56 < fjp> zinosat: I'll look at your meeting report later. 17:56 < bubulle> so, fjp's here.. 17:57 < Kamion> I'm sort of here, but lots of housework to do before leaving for a week so I'll be in and out at best 17:58 < bubulle> hmmm, not much attendants 17:59 * bubulle begins worrying 17:59 < bubulle> will we do a meeting with 2.5 people? 18:00 < bubulle> Official meeting opening hour 18:00 < bubulle> fjp: do we really maintain it? We are....you, me and a bit of Kamion 18:00 < Mithrandir> hi 18:01 < fjp> bubulle: Plenty of ppl in the channel 18:01 < waldi> hi 18:01 < bubulle> yep as usual 18:01 < bubulle> aaaaa 18:01 < bubulle> OK, let's open the meeting 18:01 < netman1> Only listening in. 18:01 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: 2.6.15 kernel status 18:02 < bubulle> So this is first topic as this is what will set the beta2 schedule 18:02 < bubulle> anyone with news? 18:02 < bubulle> I did put 2 subtopics: 18:02 < bubulle> ramdisk generator status 18:02 < bubulle> moving of kernel module .udebs under the linux-2.6 common package infrastructure 18:02 < fjp> bubulle: Where did 2nd topic come from? 18:03 < bubulle> fjp: no idea 18:03 < waldi> the kernel people 18:03 < bubulle> not mine, actually 18:03 < bubulle> but are there kernel ppl around? 18:03 < waldi> yes 18:03 < bubulle> so, any idea about a date for 2.6.15 entering testing? 18:04 < waldi> not until the udev, kernel dep loop is cleared 18:04 < bubulle> and who does this depend on? 18:05 < fjp> AIUI there will be a -4 upload sometime this weekend and that could be fasttracked to testing. 18:05 < waldi> md 18:05 < bubulle> 2.6.15-4 is the one targeted for testing? 18:05 < waldi> fjp: the bug is not yet ignored, so 18:05 < fjp> waldi: I thought vorlon had said that this is not a blocking issue for move to testing but is a blocking issue for Etch release. 18:06 < bubulle> vorlon: awaken? :-) 18:06 < makx> hello sorry for beeing late 18:06 * fjp will check with vorlon later 18:06 < bubulle> fjp: ok 18:06 < zinosat> fjp, ok, for the report:thanx 18:06 < bubulle> so we currently have no real idea of the date for 2.6.15 entering testing 18:07 < bubulle> something like this week-end+10 days seems realistic? 18:07 < waldi> yes 18:07 < bubulle> fine 18:07 < makx> seconded 18:08 < bubulle> something to add about kernel->testing? 18:08 < makx> there is an md leak which was chased on lkml 18:08 < makx> didn't check today if patch already available 18:09 < makx> but we'll get it for sure if comfirmed next days 18:09 < fjp> makx: Is that RC or could it just be fixed in a later upload? 18:09 < makx> pretty bad for md on scsi 18:09 < makx> and only affecting >= 2.6.15 18:09 < bubulle> what about the status of ramdisk generators (beware, I'm ignorant of this, so please discuss this alon, I'm just watching and taking notes..) 18:10 * joeyh wanders in rubbing sleep from eyes 18:10 < makx> klibc is progressing 18:10 < makx> svenl provided ppc fix 18:10 < makx> i need to add some compat stuff for older udev 18:10 < makx> to lower the dep of the needed udev 18:11 < makx> will do beginning next week 18:11 < bubulle> does this block us in some way? 18:11 < fjp> Is abandoning klibc an option? I understand there's a lot of stuff that does not support it so dropping libc seems unlikely ATM. 18:11 < waldi> it uses 2 or three binaries from klibc-utils 18:11 < Kamion> that's certainly the original initramfs-tools author's preferred option 18:11 < makx> well depends how busybox upstream is picking up 18:11 < Kamion> medium-term anyway 18:12 < waldi> makx: just ask 18:12 < makx> spoke with jbailey to have the klibc as hooks 18:12 < makx> still target for minimal initramfs image 18:12 < Kamion> klibc-utils could be built against glibc too - weird I know, but still 18:12 < Kamion> obviously busybox'd be better 18:12 < fjp> ramdisk generators do not block d-i IMO; we have working option for all archs now with added code in base-installer 18:13 < fjp> Although we have not seen testing results for some of them... 18:13 < makx> i'm working on boot from firewire 18:13 < makx> that's currently not supported 18:13 < bubulle> fjp: OK, so if we have a good picture now, we can maybe close it and discuss deeper in a post-beta2 meeting? 18:13 < makx> and have seen some wired timeouts on usb 18:14 < makx> that is only affecting the debian initramfs-tools as udev in ubuntu is a very different biest 18:14 < bubulle> 2nd subtopic: "moving of kernel module .udebs under the linux-2.6 common package infrastructure" 18:14 < fjp> Yes. It's up to porters to let us know which generator works best for them and then we can make sure that is default in base-installer. 18:15 < bubulle> so, this was bringed by the kernel ppl...is this relevant for beta2 or better at the end of the meeting? 18:15 < waldi> no, it is not relevant for beta2 18:15 < bubulle> OK, move to the end of the meeting then 18:15 < makx> yes mid term plan 18:15 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: Status of the udebs transition: 18:16 < bubulle> general udebs? fjp? 18:16 < fjp> Almost all udebs have now been uploaded for Beta2, both general and arch specific. 18:16 < fjp> hppa and m68k need to do some work still 18:16 < bubulle> so just missing a few more translation uploads here and there (base-installer) and that's all? 18:17 < bubulle> any hppa or m68k ppl around? 18:17 < fjp> also one more upload for base-installer (translations) and hw-detect (sparc sbus support) needed 18:17 < bubulle> maybe localechooser if I NMU iso-codes, also 18:17 * fjp will contact hppa/m68k ppl by mail 18:17 * bubulle tried to hook up ptitlouis to hppa... 18:18 < fjp> bubulle: probably also choose-mirror too? 18:18 < bubulle> fjp: yes this could bring a few more country names translations 18:18 < bubulle> but I need first to be sure about iso-codes, I'm nmu'ing Alastair packages a lot..:-) 18:19 < bubulle> fjp: what needs to be done for hppa and m68k? 18:19 < bubulle> (for the meeting report) 18:19 < fjp> So, with udebs uploaded we should now use period until they can be moved to testing for some heavy testing 18:19 < bubulle> what is the recommended testing: use sid_d-i daily builds, right? 18:19 < fjp> hppa: needs switch from 2.6.14 to 2.6.15 + upload of arch specific udebs 18:19 < joeyh> hmm, I think I've found another blocker. We need locales 2.3.5-10 in testing for curent localechooser 18:20 < bubulle> joeyh: good point, yes 18:20 < bubulle> crap, bouz is not online 18:20 < fjp> m68k: needs kernel udeb updates (also 2.4.27) and also arch specific udebs 18:21 * bubulle checks what's blocking locales 18:22 < bubulle> If I'm correct, glibc just needs a hint 18:22 < CIA-6> debian-installer: fjp * r34611 scripts/testing-summary/annotations: localechooser may need locales 2.3.5-10 18:22 < joeyh> bubulle: looks like 18:22 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: Release blockers 18:22 < bubulle> so, we covered a few of these 18:22 < bubulle> any others? 18:23 < joeyh> any news on localization-config? 18:23 * fjp is not aware of any real blockers 18:23 < bubulle> joeyh: well, markos leaves for military service in 2 weeks...:-( 18:23 < bubulle> he's trying to get a new maintainer 18:23 < fjp> markos_: You there? 18:24 < bubulle> I suggested him to post on -i18n 18:24 < joeyh> seems netinsts are still broken as per d-i today 18:24 < fjp> Should missing l-c be considered a blocker? It is a regression I guess... 18:24 < bubulle> I think that l-c is a bit less critical now as xserver does quite a good job in setting the keyboard 18:25 < bubulle> and gdm login is very well localized without l-c 18:25 < bubulle> given what I see from the babelbox tests, I would not declare l-c a blocker 18:25 < fjp> joeyh: Build or install? 18:25 < bubulle> probably document the regression in the errata 18:25 < joeyh> fjp: install 18:26 < bubulle> another issue for the errata: the input problem reported by chicky today 18:26 < bubulle> should be nice to fix, though 18:26 < fjp> joeyh: tbm tracked down an issue because of included .svn files in udebs; should be fixed tomorrow. 18:26 < bubulle> the netinsts are missing some packages, IIRC, that it? 18:26 < bubulle> that's 18:27 < joeyh> yes, missing some stuff still it seems 18:27 < bubulle> is someone working on this? 18:27 < fjp> Yes, sledge been working on it 18:28 < bubulle> sledge does not seem around in the channels I'm listening... 18:28 < bubulle> this is obviously a blocker 18:28 < fjp> Try debian-cd. 18:29 < luk> or debian-tech 18:29 < fjp> Yes, but also something that I would expect to be fixed quite easily. 18:29 < fjp> bubulle: No need to fix this during this meeting. 18:29 < bubulle> Sledge: if you OK, just tried to get his input 18:30 < bubulle> oops 18:30 < bubulle> I wanted to say "just tried to get his input" 18:30 < bubulle> something else for blockers? 18:30 < fjp> So what about UTF-8? There seem to be quite a few little regressions. 18:30 < fjp> IMO it really needs someone to coordinate this. 18:31 < bubulle> well, the input thing seems to be one (probably something we had for a long time) 18:31 < bubulle> yes, would be nice to have someone and, please, not me 18:31 < bubulle> ptitlouis mentioned me today that hppa console has some weird problems with UTF-8 18:32 < bubulle> chicky: you're the one who started a page about it...:-) 18:32 < fjp> joeyh: The tasksel problems when not in UTF-8, that was not really anticipated, was it? 18:32 < bubulle> hmm, no chicky around... 18:33 < joeyh> not entirely anyway.. 18:33 < fjp> joeyh: Would the same problems occur if langs all used their own encoding? 18:34 < fjp> Is there some other way that could be fixed? 18:34 < VivekVC> bubulle: hi, i am here. I confirm 18:34 < chicky> bubulle: hi (just comming back) 18:34 < bubulle> chicky: I was about to nominate you as UTF-8 transition coordinator..:-) 18:35 < fjp> bubulle: ask ptitlouis to file an installation report 18:35 < joeyh> one way to fix it would be to run tasksel in utf-8 locales but not necessarily the finished system 18:35 < chicky> well, I don't understand i18n too much 18:35 < bubulle> chicky: we actually mor eneed someone to track down things and nag people who have to be nagged and so on... 18:36 < bubulle> no need to understand everything...believe me I often talk about things I don't understand 18:36 < chicky> ok, I guess I could try this 18:36 < bubulle> \o/ for chicky 18:36 < bubulle> fjp: do we consider these utf-8 issues as blockers? 18:37 < bubulle> a workaround would be switching all remaining languages to UTF-8 18:37 < fjp> I would say not. The idea was to go with it for one release and see what problems we get. Then decide if we want to revert or not. 18:38 < bubulle> agreed 18:38 < bubulle> final call for other release blockers! 18:38 < fjp> That's also why we do need someone to pay attention. 18:38 < _max> hi all 18:38 < bubulle> fjp: now we have someone 18:38 < fjp> :-) 18:39 < fjp> Let's move on 18:39 < bubulle> _max: yay we'll soon talk about partman-crypto 18:39 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: Finalize the schedule 18:39 < fjp> Not all that much to say about that. 18:39 < bubulle> fjp published a nice schedule....does it need to be changed in some way? 18:39 * bubulle tries to find an URL for it 18:40 < _max> great :) 18:40 < fjp> We know it depends on kernel entering testing. I'll update the schedule at that point with real dates and some more detail 18:40 < fjp> Main activitity until then is testing. 18:41 < fjp> Example: I discovered last week that partman-auto-lvm does not work on sparc 18:41 < bubulle> fjp: but tests we can do now should be done with sid or testing installs? 18:41 < bubulle> my own testign installs often fail because of initrd problems 18:41 < fjp> sid (daily builds) of course 18:42 < bubulle> yep...and better install sid or testing? 18:42 < fjp> Does not really matter for 1st stage issues. 18:43 < bubulle> can we begin to fill in the release-checklist file now? 18:43 < joeyh> are the cd links still pointing at sarge_d-i cds? 18:43 < joeyh> er, etch_di-i 18:43 < fjp> No, daily is pointing at sid_d-i 18:44 < joeyh> ok 18:44 < fjp> (has been for a long time) 18:44 < bubulle> do we move on...nothing mor eto say it seems 18:44 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: status of partman-crypto 18:44 < fjp> bubulle: release checklist should be done when we switch to sarge_d-i I would say. 18:44 < bubulle> ok 18:45 < bubulle> _max: here you go 18:45 < fjp> i.e. after everything has been moved to testing 18:45 < _max> ok 18:45 < _max> unfortunately there has not been much progress in the last weeks 18:45 < _max> (plus i'm recovering from disk crash this very moment) 18:46 < joeyh> are the main blockers technical or maintainer not responding? 18:46 < _max> so I think the status for beta2 is obvious :) 18:46 < bubulle> IIRC, from you wiki page you depend on a few things 18:46 < fjp> _max: The main thing is external stuff, right? 18:46 < _max> joeyh: mainly maintainer (non)action 18:46 < bubulle> hmmm, NMU gnupg? :-) 18:46 < fjp> waldi: Did you reconsider the busybox patch that _max needs? 18:47 < _max> i meant to write to james about his thoughts on a gnupg-udeb, but havent found the time 18:47 < _max> will do so soon (the writing, not the NMU ;) 18:47 < bubulle> hmm, given that gnupg had no upload since October, I bet that nothing will happen 18:48 < bubulle> fjp, joeyh: can one of you add some weight to max request for a udeb? 18:48 < bubulle> or maybe Kamion... 18:48 < _max> bubulle: at the time I filed the bug, partman-crypto was still very young and not clear that it would be going anywhere. I think if I write to james again and explain the blocking nature of the bug, he might reconsider 18:48 < CIA-6> debian-installer: pi-guest * r34612 packages/po/eu.po: [l10n:eu] Update basque translation 18:49 < bubulle> waldi: about #323436? 18:49 < _max> getting seconds on the bug obviouly wouldnt hurt :) 18:49 < fjp> Problem is that I cannot judge the technical arguments which makes it hard to push for something... 18:50 < bubulle> fjp: there are no arguments actually, just nothing..:-) 18:51 < bubulle> _max: what is cdebconf-plugin-entropy? 18:51 < fjp> bubulle: waldi rejected the busybox patch; he did consider it 18:51 < bubulle> fjp: I was more thinking about the gnupg BR 18:52 < bubulle> which seems pretty not invasive 18:52 < _max> the thing about technical reservations is that it actually just matches what upstream has been doing and recommending for a long time. i wouldn't say i'm an expert in these things, but if he considers it fine, so do I 18:52 < fjp> Anyway, partman-crypto is not going to happen for Beta2, so I suggest we shelve it. 18:52 < _max> ok 18:52 < bubulle> fjp: OK...but anyway we do not have other beta2 issues..:-) 18:53 < bubulle> unless I'm wrong 18:53 < _max> i'll still be working on it and trying to get the missing bits added, perhaps in time for another release 18:53 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: g-i: summary of Extremadura activities and general point about g-i 18:53 < fjp> _max: I suggest you try to convince ppl yourself and maybe discuss the issues on d-devel if you are ready to consider NMUs 18:54 < bubulle> fjp: can you please chair the remaining of the meeting? I have to drive my daughter for 10 mins 18:54 < fjp> K. 18:54 < fjp> zinosat: Want to give a short summary of Extremadura? 18:54 < zinosat> Kamion, have you seen the patch for #339379? 18:54 < zinosat> ok 18:54 < _max> fjp: ok. I wouldn't go as far as NMUing the packages, but rather keep local patches to the udebs and make separate builds of the (crypto)installer 18:55 < zinosat> I guess we had the chance to discuss most of the pbs "live" 18:55 < fjp> zinosat: Yes, Kamion is OK with the patch. 18:55 < zinosat> we fixed the vt switch bug (which IMO was the main blocker for g-i) 18:56 < zinosat> got upstream ppl to do most of the things we needed: 18:56 < zinosat> push gdk-directfb backend into mainstream (still to be done but we had an OK) 18:57 < zinosat> we found a terminal usable in g-i (dfbterm) and the author is ok submitting patches/improvements 18:57 < fjp> For me, most important was to get to meet everybody and especially upstream ppl. 18:58 < zinosat> yes, and Mike is following debian-boot 18:58 < fjp> A lot was discussed and started; some bugs were solved; we've now got business card CD image as well as mini.iso. 18:58 < zinosat> I expected more on the status of fonts 18:59 < fjp> Extremadura as a meeting place is excellent: good facilities, excellent food. 18:59 < joeyh> speaking of g-i, the dpkg-dev fix goes into unstable today, so will a debhelper to use it 18:59 < fjp> joeyh: Great. Was going to ask about that. 18:59 < zinosat> cool! 19:00 < fjp> That means we can start the slow update of libs to use it. I will start filing bugs with patches on that. 19:00 < joeyh> libd-i would be a very nice start 19:00 < fjp> zlib should be one of the first... 19:00 < joeyh> then daily builds would stop breaking due to it being out of date on the build system iirc 19:00 < fjp> Ah, yes. That would be nice. 19:01 < zinosat> I think now we need to make g-i speak all languages. I can start working on it next week. mailed the Persian guy with a testing iso a few minutes ago 19:01 * bubulle back 19:02 < bubulle> fjp: do we need the fixed dpkg in testing? 19:02 < zinosat> once we have daily builds and fonts fixed g-i will be in a really good shape 19:02 < joeyh> not especially 19:02 < fjp> bubulle: No, not really. 19:02 < bubulle> ok 19:03 < bubulle> because I suspect it will need some time to cool down..:) 19:03 < fjp> Only thing I'm worried about is loads of uploads delaying migration. 19:03 < bubulle> I guess that a report of the overall Extremadura meeting will happen? 19:03 < fjp> For g-i stuff to move to testing the freetype RC bug is a major blocker 19:03 < zinosat> I wrote a draft 19:04 < fjp> Yes. 19:04 < bubulle> anyway, I personnally want to thank zinosat for organizing it so uqickly and so well... 19:04 < bubulle> this was certainly a huge push for g-i 19:04 < zinosat> it's been a pleasure, really 19:04 < zinosat> would be nice if we could do an annual g-i thing 19:05 < fjp> bubulle: Move to last subject: kernel udebs? 19:05 < bubulle> fjp: which RC bug? 19:05 < joeyh> it seems trivial to fix 19:05 -!- bubulle changed the topic of #debian-boot to: D-I development team meeting in progress | Agenda: http://wiki.debian.org/DebianInstallerMeetings | Current topic: kernel udebs 19:05 < fjp> bubulle: All of them really; basically unversioned API changes 19:06 < fjp> Well, makx has wandered off. 19:06 < bubulle> kernel people?? 19:06 < fjp> waldi: ? 19:08 < bubulle> OK, this seems to close the topic...and the meeting 19:09 < bubulle> thanks you to all people who attended and work hard for beta2 now.... 19:09 < bubulle> off topic discussion and post-meeting chat can begin...