[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

[MEETINGS] Log from the Debian Installer team meeting of June 24th 2006

Please find below the log from the Debian Installer "monthly meeting
that's held every two months" of June 24th 2006 on the #debian-boot
IRC channelm on irc.debian.org

Log from the Debian Installer team meeting of June 24th 2006, 16:00UTC

People who speak below:

Alphix    : David Härdeman
bubulle_  : Christian Perrier
dajobe    : Dave Beckett
fjp       : Frans Pop
h01ger    : Holger Levsen
joeyh     : Joey Hess
jvw       : Jeroen van Wolfelaar
maks      : Maximilian Attems
_Max      : Max Vozeler
svenl     : Sven Luther
trave11er : Jurij Smakov
waldi     : Bastian Blank
zinosat   : Davide Viti

Nicknames mentioned during the meeting though these people were not attending:

ninou     : Sylvain Ferriol 
vorlon    : Steve Langasek

Hours are European Time (UTC +0200):

17:57 < bubulle_> OK, time for people who are actually around to say "hi"
17:58 < waldi> hi
17:58 < Alphix> hi
17:58 < h01ger> hi
17:58 < maks> hi
17:58 < zinosat> hi
17:58 < _Max> ho
17:58 < joeyh> "hi"
17:58 < trave11er> oh, it's a meeting, isn't it?
17:58 < CIA-1> debian-installer: bubulle * r38402 installer/doc/i18n/en/new-language.xml: 
17:58 < CIA-1> debian-installer: Mention that needed-characters is required only for languages that are
17:58 < CIA-1> debian-installer: supported by the next interface
17:59 <+fjp> bye
17:59 < svenl> fjp: excpect the whole set of patches for monday, *
17:59 -!- mode/#debian-boot [+o waldi] by ChanServ
17:59 -!- mode/#debian-boot [+b *!*Kacak-4-@81.214.40.*] by waldi
17:59 -!- _TURK_ was kicked from #debian-boot by waldi [this is too much scripting for developers]
17:59 < svenl> bubulle_: hi
17:59 -!- mode/#debian-boot [-o waldi] by waldi
17:59 < bubulle_> Reminder: the meeting agenda is http://wiki.debian.org/DebianInstaller/Meetings
17:59 < bubulle_> it's a bit messy this time..:-)
18:00 < CIA-1> debian-installer: fjp * r38403 installer/ (18 files in 2 dirs): 
18:00 < CIA-1> debian-installer: - Remove needed_characters files for "level 4" languages (are not supported in newt frontend)
18:00 < CIA-1> debian-installer: - Rename needed_characters files for prospective languages so they're not included yet
18:00 < bubulle_> well, time to open the meeting, folks
18:00 < bubulle_> will be pretty dense so, let's go for the first topics
18:00 < bubulle_> Status reports 
18:00 < bubulle_> [WWW] Partman crypto - 10 mins
18:00 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - topic partman crypto
18:01 < bubulle_> http://lists.debian.org/debian-boot/2006/06/msg00893.html
18:01 <+fjp> _Max or Alphix: here?
18:01 < _Max> ok, I'll be brief as I'm having a headache :-/
18:01 < Alphix> The first point of that mail is no longer relevant...partman-crypto 6 has been uploaded
18:02 < Alphix> partman-auto-crypto is at the legendary state works-for-me (tm)
18:02 < Alphix> most other udeb's that partman-crypto depend on seem to be in good shape
18:02 < bubulle_> Alphix: from your report, I would say that putting it in beta3 could be the only opportunity to give it mor eexposure
18:02 < bubulle_> it==partman-auto-crypto
18:03 < Alphix> perhaps...there was little feedback gained from posting to debian-devel...a decision for the RM
18:03 <+fjp> I'm not sure that p-a-c is ready yet for beta3.
18:03 < _Max> I've started to review it a bit .. will send feedback to the list later
18:03 < bubulle_> fjp: if it's not in beta3, do you think it can be in the release?
18:04 < Alphix> It duplicates a lot of code from partman-lvm, if the default was changed for partman-lvm to put swap partition on lvm, the amount of code duplicated could be reduced
18:04 <+fjp> I've not yet had time to review. I'd suggest having just partman-crypto in beta3 and adding p-a-c for RC1.
18:04 < Alphix> I've posted to debian-boot and debian-devel on that issue, but received little feedback on it as well
18:04 < Alphix> as for the actual code of partman-crypto, I think testing and documentation are the two main issues
18:05 <+fjp> What we really need for partman-crypto is some documentation. Alphix and _Max have promised to work on that.
18:05 < _Max> agreed
18:05 < bubulle_> fjp: do you make this a requirement for being in beta3?
18:05 < waldi> fjp: i intend to fix the module problem in the next days, so it can be integrated into the d-i kernel build after beta3
18:05 < Alphix> And there is the problem with root-on-crypto...the password prompt has no keymap loaded in the initramfs image
18:05  * fjp does not understand
18:06 < Alphix> That's #337663 which has been marked wontfix
18:06 < maks> en-de keymap
18:06 < _Max> see partman-crypto/TODO
18:06 < maks> Alphix: for general initramfs it is not relevant
18:06 < waldi> Alphix: the problem is, which keymap to integrate?
18:06 < Alphix> in the initramfs image...when you get a prompt to enter the password (when booting your newly installed system), no keymap is loaded....not sure what to do
18:06 < Alphix> maks: agreed
18:07 < Alphix> And I should add that I haven't really alerted maks to the issue yet :)
18:07 <+fjp> Ah. Not really an installer problem, but a usability problem on reboots.
18:07 < maks> cryptsetup can add keymap, but should wisely choose..
18:07 < bubulle_> Alphix: I think that if that's properly documented, people can live with having to enter the pwd on a US-mapped kbd
18:08 < bubulle_> and, yes, that's me saying this...the user of the crappy French kbd
18:08 < Alphix> bubulle_: yes, I'll probably add a warning to the "enter password" template as a temporary solution
18:08 < bubulle_> anything else on p-c?
18:08 <+fjp> Could we detect this in d-i and add a keymap in the initramfs config when needed?
18:09 < Alphix> fjp: probably....I'll look into a fix for cryptsetup initramfs scripts
18:09 < maks> only if that hook gets added to regular cryptsetup too
18:09 <+fjp> Or is there no support for keymaps at all currently in initrd generators?
18:09 < Alphix> none at all in initramfs-tools as far as I know
18:09 < maks> correct
18:09 < bubulle_> if that's done, I suggest someone considers the BR to grub-installer to make it use an appropriate keymap depending on the current keymap
18:10 <+fjp> OK. I'll leave you to sort that out. Is a documentation issue at the very least.
18:10 < bubulle_> waldi: juhu..:)
18:10 < Alphix> but grub-installer would be a completely separate issue....
18:10 < waldi> bubulle_: yep :)
18:10 < maks> Alphix not from a user point of view
18:10 < Alphix> agreed
18:11 < _Max> question: should we make dm-crypt the default?
18:11 < _Max> I think that would be good
18:11 < maks> what is current default?
18:11 < bubulle_> OK, we're out of time for p-c folks
18:11 < Alphix> _Max: I agree
18:11 < bubulle_> more to add?
18:11 < Alphix> maks: loop-AES
18:11 < maks> ooh yes dm-crypt
18:11 < bubulle_> the conclusion seems: p-c in beta3 and no p-a-c
18:12 < waldi> bubulle_: and dm-crypt default
18:12 < Alphix> bubulle_: I've got nothing more to report at least
18:12 < _Max> ok, let's do it. we'll sort out the details ;)
18:12 <+fjp> Alphix, _Max: please concentrate on getting p-c release ready.
18:12 < _Max> fjp: will do
18:12 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Graphical installer 10 mins
18:12 < bubulle_> http://lists.debian.org/debian-boot/2006/06/msg00902.html
18:12 < zinosat> fonts and new libraries are still the hot topics for g-i. 
18:12 < Alphix> I'll be somewhat absent minded from now on....world cup :)
18:12 < zinosat> cairo is avilable as experimental set of packages and Joss said he would have tried to work on gtk+-directfb during this weekend
18:13 <+fjp> Alphix: Why? You've already lost AFAICT from radio...
18:13 < zinosat> this on the libs side
18:13 < bubulle_> anyone could try to bring Joss in right now?
18:13 < Alphix> fjp: don't torment me
18:13 < dajobe> I'm going to make a new cairo directfb exp deb today
18:13 < zinosat> ttf-thai-udeb made it in the archives: ttf-bpg-georgian-fonts, ttf-farsiweb, ttf-tamil-fonts still missing
18:13 < zinosat> dajobe, what about the cairo-directfb-dev.deb
18:14 < dajobe> zinosat: that's in the package too
18:14 < bubulle_> jvw: around?
18:14 <+fjp> There's still some uncertainty about the libs. Hopefully dajobe, Joss and attilio/davide can sort it out while working on it
18:14 < jvw> bubulle_: sí
18:14 < zinosat> cool. I think it makes alot of sense to include it
18:14 < bubulle_> jvw: could something be done for those package mentioned above, which need NEW processing because they have a udeb added?
18:14 < jvw> notlistening/attending this meeting though, so whatever your question was, please repeat?
18:15 <+fjp> Important is that nothing is uploaded into unstable until all is ready as else we'd break existing g-i.
18:15 <+fjp> Uploading to experimental is fine, but experimental does not support udebs yet...
18:15 < dajobe> for cairo, it won't go into unstable till 1.2.0 is released
18:15 < zinosat> in anutshell: it's possible to use directfb in place of X, so that deb could be useful for ppl choosing that solution
18:15 < waldi> fjp: AFAIK it does
18:15 < waldi> fjp: but the PAckages files are not generated
18:15 < jvw> I'm happy to process some packages through NEW if there's some need for them etc -- mostly, if fjp requests such things, I do it right away
18:15 <+fjp> waldi: OK
18:15 < jvw> because I know fjp doesn't request so lightly
18:16 < zinosat> fjp, I'll write how to "migrate" to the new libs in the wiki
18:16 <+fjp> zinosat: Did you mean cairo-directfb-_dev_.deb or a normal lib package?
18:16 < zinosat> the first
18:17 <+fjp> OK
18:17 <+fjp> zinosat: Do we still want fc-list?
18:17 < zinosat> good point. admit I haven't dug into that
18:17 < zinosat> for now no.
18:18 < bubulle_> zinosat: what about this #374902 which seems to hurt g-i?
18:18 < svenl> BTW, what of [-oskar-]'s C reimplementation of gparted ? He ended his 'stage' yesterday, i am not sure how fully mature it is, but it could benefit from widder exposure, and future development in case [-oskar-] has no time to follow on it.
18:18 <+fjp> We need something for font switching though.
18:18 < CIA-1> debian-installer: adn-guest * r38404 packages/po/ar.po: [l10n] Synchronise with Arabeyes CVS
18:18 < svenl> the code is in the parted svn on alioth.
18:18 <+fjp> svenl: To be honest, we have not seen anything from oskar...
18:18 < zinosat> bubulle_, it would be great if vorlon uploaded a package with the fix
18:19 < bubulle_> well anything more on g-i?
18:19 <+fjp> I doubt it's an option for Etch anyway, but if someone wants to adopt it that would be fine.
18:19 < zinosat> fjp, switching font is needed, fontsize change should be handled by fixing the ttf file in the udeb IMO
18:20 < svenl> fjp: he is waiting for the 2.8+ gtk stuff, so no wonder.
18:20 <+fjp> Oskar should do an extensive report on status to d-boot.
18:20 < zinosat> bubulle_, I'm about to file a bug against freefont
18:20 < bubulle_> zinosat: bleh..:)
18:20 < bubulle_> OK, let's move
18:20 <+fjp> zinosat: Let's discuss that first...
18:20 <+fjp> bubulle_: Wait.
18:20 < bubulle_> ok
18:21 <+fjp> Anything for Np239 ?
18:21 < svenl> fjp: i am just telling this here so people are aware, and know that if he doesn't follow on this now, that someone other interested wants to jump in
18:21 < zinosat> Np239, it'd be great if gtk+-directfb would be available
18:21 < bubulle_> fjp: I tried to ping Np239 but he seems away right now
18:22 <+fjp> Cool. dajobe: could you upload yor next version there as well?
18:22 < dajobe> Np239: pls wait till I make the cairo 1.1.10 exp debs since some needed fixes are in there
18:22 < zinosat> Np239, if you come up with something experimental let us now (d-boot) and I'll test it
18:22 < dajobe> fjp: yes, within the hour I hope
18:22 < h01ger> svenl, a summary in mail from oskar would be good/better...
18:22 <+fjp> I'll bug ftp-masters about getting packages files for udebs there.
18:22 < zinosat> dajobe, as far as compilation is concerned 1.1.6 would be enough
18:22 < bubulle_> more stuff or can we move on now (/me looks watch)?
18:23 <+fjp> Move on I think.
18:23 <+fjp> dajobe, Np239: thanks for your work
18:23 < zinosat> yeah
18:23 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Beta 3 release - Main blocker is that we need 2.6.16 in testing
18:23 < bubulle_> Reminder: overview is http://wiki.debian.org/DebianInstaller/EtchBeta3Prep
18:24 <+fjp> Release preparations have basically just really started.
18:24 <+fjp> Tentative plan is release in 3 weeks.
18:24  * bubulle_ would really like to see it before leaving so 3 weeks is fine
18:25 <+fjp> Depends on what happens with 2.6.16 kernel migration to testing. If that goes normally, three weeks should be doable.
18:25 < bubulle_> About 2.6.16, what are we missing?
18:25 < waldi> meta packages
18:25 <+fjp> And kernel udebs for some arches.
18:25 < bubulle_> which?
18:26 <+fjp> Currently most. I'll start bugging porters after the weekend.
18:26 <+fjp> waldi: I did s390 yesterday (but you probably saw that)
18:26 < waldi> fjp: yep
18:26 < bubulle_> do we have visibility on the kernel release we will ship Etch with?
18:27 <+fjp> Not really.
18:27 < maks> not decided yet afaik
18:27 < waldi> bubulle_: no
18:27 <+fjp> Consensus!
18:27 < bubulle_> :)
18:27 < maks> 2.6.17 has arch advantages
18:27 < maks> but not yet many stable releases on his back ;)
18:27 <+fjp> Any questions or issues relevant for the release?
18:28 < bubulle_> OK, more about kernel release (with beta3 in mind)?
18:28 < bubulle_> fjp: twinned brains...
18:28 < waldi> no
18:28 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Beta 3 release - Massive coordinated migration of udebs needed: most interdepend
18:29 < bubulle_> fjp, joeyh: this is usually your baby..:)
18:29 <+fjp> Not much to say about that. We have a lot of structural changes where udebs depend on each other.
18:29 < joeyh> yeah, I think this is best approached by checking if there are any updated udebs we don't want to let in
18:29 <+fjp> So basically, all udebs have to migrate together and beta2 will break as soon as we do.
18:30 < bubulle_> fjp: so it's likely that at some moment, migrations break testing installs..:-)
18:30 < joeyh> note that there are also several removals that will need coordination, eg prebaseconfig
18:30 <+fjp> joeyh: I don't know of any.
18:30 < waldi> fjp: partitioner, partconf?
18:31 < bubulle_> fjp, joeyh: most uploads to unstable have already been done, right?
18:31 < joeyh> we also need to check if any non-di udebs will be held out by any other issues with their debs
18:31 <+fjp> What I do want to ask is to check changes in trunk that could be unwanted for beta3 with me before you commit them
18:31 < joeyh> probably another round of translation uploads
18:31 < bubulle_> yeah, fjp recent mail triggered a lot of translation and QA activity
18:31 <+fjp> As joey says. There are very few pending functional changes.
18:32 < bubulle_> on that matter, while working on seppy scripts, I try to get translation status for testing
18:32 <+fjp> joeyh: Yes, I'll start checking non-d-i udebs next week
18:32 < bubulle_> this will help just before the release, to request translation uploads of non DI packages
18:33 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Beta 3 release - Some changes in debian-cd needed after migration
18:34 < joeyh> what are those?
18:34 < bubulle_> fjp: ?
18:34 <+fjp> Mostly S/390 and sparc update
18:34 < svenl> maybe some powerpc stuff too ? 
18:35 <+fjp> Also I'm thinking if we should add yaird to netinsts now that size is relaxed.
18:36 <+fjp> svenl: g-i integration needs doing for ppc. I'm not aware of anything else. I'll ask colin to check things.
18:36 <+fjp> svenl: Please mail if you know about needed changes.
18:37 < bubulle_> maybe possible to briefly mention changes here so that I can mention them in the report
18:37 <+fjp> joeyh: What do you think about adding yaird?
18:37 < h01ger> fjp, g-i on powerpc is a different initrd atm?
18:37 < svenl> fjp: did the hang-at-boot on g-i/powerpc get solved ?
18:37 <+fjp> h01ger: Currently only a separate mini.iso is available
18:38  * h01ger will get a oldword g3 tonite - hopefully g-i testing on that machine is feasable and fun :)
18:38 < joeyh> well, some mixed feelings
18:38 <+fjp> svenl: No idea. I think eddyp still had problems.
18:38 < h01ger> fjp, ah. cds dont boot on oldworld... (our cds..)
18:38 < maks> fjp: yaird upstream seems mia
18:38 <+fjp> maks: Yes, I know.
18:39 <+fjp> svenl: Did you used to build g-i for oldworld?
18:39 < maks> please take a look for the table on the debian wiki
18:39 < maks> http://wiki.debian.org/InitrdReplacementOptions
18:39 < Alphix> maybe I should add that partman-crypto has had no testing as far as I know with yaird
18:40 <+fjp> initramfs-tools will remain the default
18:40 < svenl> fjp: i never tried on oldworld, the same image should work just fine, the problem is with the ramdisk size, which may hit some oldworld firmware size limits.
18:40 < svenl> fjp: naturally, doing miboot-floppies g-i strikes me as less than optimal, so the bootx method is the way to go.
18:40 < maks> fjp: so for what advantage?
18:40 < svenl> fjp: i know g-i doesn't work on prep.
18:41 <+fjp> maks: Choice mostly.
18:41 < svenl> fjp, h01ger: maybe post beta3, we could imagine a two-step g-i thingy, where the additional code is not part of the ramdisk, but pulled in later on.
18:42 < zinosat> h01ger, if you plan to look at the g-i crash on ppc ping me next week
18:42 < bubulle_> OK, we're running out of topic right now..:-)
18:42  * fjp feels g-i on oldworld is probably low priority
18:42 < bubulle_> this is probably "Ports need testing!"
18:42 < h01ger> zinosat, i doubt i'll have time next week (before the weekend). and if, i'll look into general oldworld issues first
18:43 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Beta 3 release - Ports need testing!
18:43 < svenl> fjp: agreed
18:43 <+fjp> Well, all we can do is ask...
18:43 < zinosat> h01ger, np
18:43 < bubulle_> fjp: which ports do you think are currently loosely tested?
18:43 <+fjp> h01ger: Agree that is prio.
18:43 < waldi> fjp: send a mail to d-d-a?
18:44 <+fjp> i386, amd64 (less), s/390 (good)
18:44 <+fjp> Don't know about others.
18:44 < svenl> i will do tests on 64bit apple (Xserve G5), pegasos, and PReP, and get a remote test done on IBM JS21 blades and power5 p505, and maybe on the quad power5 of augsbourg.
18:44 <+fjp> Will test sparc and probably  hppa myself soonish
18:44 < bubulle_> alpha?
18:44 < bubulle_> m68k?
18:44  * joeyh hopes to get the dilab running again in a week or so, taggart got it a replacement switch
18:44 <+fjp> waldi: Yes, good idea.
18:44 < svenl> this leaves 32bit powermac, and oldworld, and 32bit IBM chrp/rs6k, apus is currently not supported.
18:45 <+fjp> svenl: Good.
18:45  * zinosat needs to go out: bye everyone
18:45 < h01ger> svenl, 32bit powermac=newworld?
18:45 < waldi> h01ger: yes
18:45 < svenl> h01ger: yeah, even though it rejoined the not-produced-anymore arches :)
18:46  * h01ger might be able to do this next weekend..
18:46 < maks> jay estabrook from hp tried several alphas and had troubles ->http://lists.debian.org/debian-alpha/2006/06/msg00034.html
18:47 <+fjp> Hopefully vorlon can put some time into alpha.
18:47 < bubulle_> fjp: agree with waldi's suggestion to a mail in d-d-a for ports testing?
18:47 <+fjp> [18:44:56] <fjp> waldi: Yes, good idea.
18:47 < bubulle_> sorry
18:47 < bubulle_> OK, let's move if you don't mind
18:48 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Beta 3 release -Known issues not listed on overview page? 
18:48 < bubulle_> I guess this is an opened question by fjp. Who's aware of such things?
18:48 <+fjp> Mail to d-boot is prefered though. h01ger's overview was a nice example.
18:49 <+fjp> If nothing, we can move on.
18:49 < bubulle_> well...
18:49 < h01ger> from there: miboot-floppies cannot be part of beta3. but i'd like to mention that somewhere...
18:49 < h01ger> atm the beta2 floppies link to non-working ones..
18:49 < bubulle_> what about the work to make floppy builds work again?
18:50 <+fjp> h01ger: I'll ask Yoe to build those images too, just like Colin did.
18:50 < joeyh> IMHO we should stop building non-miboot floppies that don't work. It's just cruel
18:50 < h01ger> joeyh, ack
18:50 < svenl> bubulle_: needs someone to followup and coordinate. its on good way.
18:50 < h01ger> colin has working builds, just at a slightly different url...
18:50 < waldi> YEAH!
18:50 < Alphix> BAH!
18:50 < bubulle_> waldi is ack online now
18:50 <+fjp> joeyh: Could you look into ppc floppy size with h01ger?
18:51 < joeyh> guess so
18:51 <+fjp> AIUI net-drivers is too big currently.
18:51 < bubulle_> ok, let's move....
18:51 < h01ger> yup
18:51 < maks> i've an klibc that adds mknod (easy) need to look into an load_initrd for floppy
18:51 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - ppc64 cross||native installer
18:51 <+fjp> Hopefully Colin will have a bit more time starting next week too.
18:51 < bubulle_> most issues are detailed on http://wiki.debian.org/DebianInstaller/Meetings
18:52  * Alphix needs to go....
18:52 < bubulle_> Alphix: thanks for participating...
18:52 < maks> ppc64 was already discussed afair 2 meeting ago
18:52 < svenl> i have something to say about the ppc64 port.
18:52 < maks> with not much enthusiasm
18:52 < svenl> I believe, as i said on list, that the best solution on this is to do the cross-install thingy.
18:53 <+fjp> svenl: I think your view is well known from your list postings.
18:53 < svenl> multiarch for etch is as good as dead, and so ppc64 effort may be worthwhile to consider until we have multirch-post-etch.
18:53 < svenl> but doing a 64bit installer doesn't bring anything.
18:53 <+fjp> The main question for us is if there is any harm in supporting it with the proposed conditions I listed in the wiki.
18:54 <+fjp> joeyh: ?
18:54 < bubulle_> I think that the general question is what's under "proposed conditions" on the meeting agenda
18:54 < joeyh> I agree with those conditions
18:54 < svenl> as afor the patches, the multiaech effort will use powerpc64.
18:54 < bubulle_> yikes, twinned brains again
18:54 <+fjp> joeyh: So, allow it based on those conditions?
18:55 < bubulle_> seems to be some kind of general consensus
18:55 <+fjp> I don't see much harm in it myself.
18:55  * joeyh neither
18:55 < svenl> but i would mention an effort to do the cross-install thingy post-beta3, if possible.
18:55 < svenl> me neither, the only possible harm is not going for the real solutions afterward :)
18:55 < bubulle_> svenl: does such effort follow the "proposed conditions"?
18:56 <+fjp> svenl: I agree that it does not bring anything for Debian as a whole, but that does not mean that we need to deny Andreas' request. He's been very friendly about it.
18:56 <+fjp> svenl: I think that will probably be post-Etch. Preparations can start earlier of course.
18:57 < svenl> fjp: well, the trouble i have with Andreas, is that he does his one-man-thingy, approaches DDs with official 64bit porter hat, but was not present when asked to participate in a multiarch effort.
18:57 < bubulle_> move to next topic...seems to be a good occasion...:-)
18:57 <+fjp> Yep.
18:57 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Post-Beta3 development
18:57 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Post-Beta3 development - Removal of 2.4 kernel support
18:58 < waldi> just do it *hide*
18:58 < maks> great
18:58 < svenl> post-beta3, can i add the ibm power stuff here, even though it is not on the agenda ? As well as some plans with a screen udeb ? at the end of the topics, if there is time ?
18:58 < bubulle_> svenl: yep, after the other topics please
18:58 < h01ger> what other "options" than "just do it" are there? :)
18:59 < bubulle_> so, so we announce to the whole world that we will do it after beta3?
18:59 < svenl> 2.4 kernel. I think it is sane to say that the kernel team will not support any 2.4 kernel for etch, and even h0lger's effort in that direction didn't go very far.
18:59 <+fjp> That's probably the main issue. We need to get all ports still using 2.4 to fully switch ASAP.
18:59 < svenl> indeed.
18:59 <+fjp> Yes, we will announce that.
18:59 < waldi> fjp: which of them uses 2.4?
18:59 < svenl> m68k may need 2.2/2.4 on some subarches.
18:59 <+fjp> mipsen and arm and m68k
18:59 < svenl> x86 floppies also, i think ? 
19:00 < bubulle_> fjp: anyway, if we drop support for 2.4 and ports do not switch, they disqualify themselves for the release, right?
19:00 < waldi> svenl: x86 is clean I think
19:00 <+fjp> m68k have already promised to fork if needed and keep support outside trunk
19:00 <+fjp> bubulle_: As I understand it, yes.
19:01 < svenl> bubulle_: issues are not as clean, no arch is fully non-2.6 compatible, only certain subarches are affected.
19:01 <+fjp> For i386 main issue is floppies, but hopefully ninou's work will be usable soon.
19:01 < waldi> fjp: they are too large?
19:01 < bubulle_> yep
19:01 < h01ger> x86 has no boot+root floppies built?!: http://people.debian.org/~joeyh/d-i/images/daily/floppy/ is almost empty.. and no build log either... :(
19:01 < joeyh> 2.4 floppies are also too large and so disabled already
19:01 < waldi> sparc also references 2.4
19:01 < h01ger> ah
19:02 <+fjp> sparc will drop 2.4 before Beta3
19:02 < maks> klibc work for boot floppies is on the way
19:02 < waldi> fjp: OK
19:02 < maks> ninou has inofficial ones irc
19:02 < bubulle_> well, in short, saying that we drop support for 2.4 is kind of an ultimatum...but I think this is one that's worth it
19:02 < maks> klibc has not yet the separete bin for load_ramdisk thing
19:02 < waldi> bubulle_: yep, please do
19:02 < svenl> does this mean we remove the 2.4 kernel .udebs thingies from the svn repo ?
19:02 < bubulle_> especially if the release team shares that opinion
19:03 <+fjp> One thing we will need to decide is how much we want to clean up components like hw-detect.
19:03 <+fjp> We'll need to discuss that soon after Beta3.
19:03 < waldi> bubulle_: i thought they will follow the kernel team which dropped support some time ago
19:03 < svenl> fjp: i think the previous decision was to leave 2.4 support in for innoficial builds, until etch is released.
19:03 < bubulle_> fjp: I would say that we coul dbe short on time to clean them up but that shouldn't harm that much...
19:03 < svenl> fjp: doing otherwise may make porters rethink their support.
19:03  * h01ger also thinks an ultimatum is worth it. otherwise we might end up with etch with 2.4.27 another time...
19:04 < maks> nah postetch glibc won't support 2.4
19:04 < bubulle_> we will be the Bad Boys...:)
19:04 < svenl> h01ger: agreed.
19:04 <+fjp> IMO we need to drop 2.4 install support in first RC. What others do unofficially is their choice.
19:04 < bubulle_> well, it seems that writing that "the whole D-I team has decided to drop support for 2.4" after beta"? ie *for Etchê would be true
19:05 <+fjp> waldi: That means 2.4 kernel images would need to stay available untill after d-i RC1.
19:05 < svenl> fjp: agreed, but cleanup will break even unofficial builds.
19:05 < waldi> bubulle_: hmm, the m68k people are no part of the d-i team?
19:05 < waldi> fjp: yep
19:05 < svenl> bubulle_: the kernel team has said so twice or so in the past already :)
19:05 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Post-Beta3 development - Persistent device naming
19:05 <+fjp> waldi: m68k people are agreed on that too
19:05 < waldi> fjp: okay
19:06 <+fjp> We seem to have persistent device naming for network devices.
19:06 < waldi> fjp: yes, except s390
19:06 <+fjp> Except for open question from joeyh.
19:06 <+fjp> joeyh: Did you look into that further?
19:07 <+fjp> Propagation from d-i to installed system>
19:07 < joeyh> oh yeah on whether d-i udev propigates the names to installed udev?
19:07 < joeyh> no, I have not but I don't see what the mechanism would be for it happening
19:07 < waldi> fjp: we should drop the devfs rules from udev
19:07 <+fjp> waldi: Do you know where the info is stored?
19:08 < waldi> fjp: /etc/udev/rules.d/z25*
19:08 <+fjp> Hmm. Does udev save previously assigned names there?
19:08 < joeyh> yes
19:08 < waldi> yes
19:08 < maks> initramfs picks /etc/udev
19:09 <+fjp> So we probably need to let post-base-installer.d copy that to installed system?
19:09 < waldi> yes
19:09 < waldi> just make the udev udeb do that
19:10 <+fjp> maybe it already does...
19:10 < maks> need to go, will read backlog later.
19:11 < joeyh> actually, I think udev-udeb is also missing /lib/udev/write_net_rules
19:11 < waldi> uh
19:11 < joeyh> and write_cd_rules
19:11 < waldi> that have to be fixed
19:12 < bubulle_> OK, let's move...we have 2-3 more topics yet
19:12 -!- bubulle_ changed the topic of #debian-boot to: D-I Meeting in progress - Post-Beta3 development - Supplemental repository support for udebs
19:12 < waldi> not yet working
19:12 < bubulle_> the next topic will be cancelled: we dealt with it already, mostly
19:12 < waldi> we may do the libdi abi bump shortly after beta3
19:12 < bubulle_> waldi: will this bea ready for RC1?
19:12 < joeyh> and /lib/udev/hotplug.functions..
19:12 < svenl> waldi: what is needed yet ? 
19:12 < waldi> bubulle_: hope so
19:13 < joeyh> is the code written somewhere already?
19:13 <+fjp> waldi: Not sure about dropping devfs as too much still uses that. Could be part of "2.4 cleanup" though but would probably prevent having support for installing Sarge.
19:13 < waldi> fjp: not really. partman does not need it
19:14 <+fjp> A lot (bootloader installers etc) assumes it
19:14 < bubulle_> ? Nothing more on the topic?
19:14 <+fjp> Persistent device naming for sdisks seems stalled
19:14 < waldi> bubulle_: yep
19:14 <+fjp> I don't really have a good feeling about that currently.
19:14 < waldi> sdisks?
19:15 <+fjp> disks
19:15 < waldi> partman needs it, needs to save the information
19:15 < waldi> the bootloader installers can use the information and should have no need for the devfs names
19:15 <+fjp> Why is it an issue for partman? Would say it's more an fstab and bootloader issue.
19:16 < svenl> I have been doing serial installs on IBM powerpc64 bit hardware. There are some issues broken there, and i will submit patches on monday. The one point i saw that was really problematic, was the absense of a second console, so i wondered if post beta3, a screen .udeb could be possible. Was this already investigated in the past and rejected or something ? 
19:16  * joeyh needs to run, see you
19:16 < waldi> fjp: it knows the information and throws it away
19:16 < waldi> fjp: so the bootloaders needs to detect them again
19:17 <+fjp> I've looked in /dev/disks, but info there looks very incomplete to me. IMO it will at least be a problem for devices that don't have sysfs support yet.
19:17 < waldi> fjp: scsi, ide, dasd have them
19:17 < waldi> what is missing?
19:17 < waldi> md needs some fixing
19:17 <+fjp> sbus
19:17 <+fjp> ?
19:18 < waldi> no proper sysfs support or missing support in udev?
19:18 < waldi> i think we should stop
19:18 <+fjp> I think my old megaraid scsi is missing.
19:18 <+fjp> Yes.
19:18 <+fjp> As I said, I don't have a really good feeling about it.
19:19 < bubulle_> OK....all this needs further discussion post-meeting or in the list
19:19 < bubulle_> I'll mention the questions of our RM about presistent naming for disks
19:19 < bubulle_> svenl: you have a GO for your additionnal topics
19:19 < waldi> (s390 will be glad to have it)
19:20 < svenl> I have been doing serial installs on IBM powerpc64 bit hardware. There are some issues broken there, and i will submit patches on monday. The one point i saw that was really problematic, was the absense of a second console, so i wondered if post beta3, a screen .udeb could be possible. Was this already investigated in the past and rejected or something ? 
19:20 < svenl> (sorry for earlier).
19:20 <+fjp> svenl: A screen udeb on same basis as open-ssh-client udeb sounds like a nice option.
19:20 < svenl> fjp: can you give more details ?
19:21 < bubulle_> svenl: I don't remember about discussions about this in the past
19:21 <+fjp> Not now and here. But you know how open-ssh-client works, right?
19:21 < bubulle_> my first impression is, yes, that this sounds like an interesting feature to have
19:21 < svenl> i have some vague memory of someone already proposing it a couple of years a go.
19:21 <+fjp> I'm not sure why you'd really need screen if you already have ssh though.
19:22 < bubulle_> svenl: mymemory may be wrong, anyway..:)
19:22 < svenl> fjp: yeah, but i don't see how it could be similar. 
19:22 <+fjp> svenl: Send mail to list with how you think it would work...
19:22 < svenl> fjp: to access the log and consoles before network is up, which is why i don't think the ssh analogy is ok, but i may misunderstand.
19:23 < bubulle_> svenl also mentioned me that it's too bad not having translations during serial installs
19:23 < svenl> fjp: ok, or we can discuss post-meeting here.
19:23 <+fjp> Please explain how screen would help there. /me does not understand that yet.
19:23 < svenl> actually, the problem is that at least in appareance, a serial install forces a english localization.
19:23 <+fjp> And we already have shell access from the menu...
19:24 < h01ger> arent that 2 topics?
19:24 < bubulle_> h01ger: yes, they are
19:24 < svenl> fjp: screen allows to have multiple consoles, so you put it in the ramdisk (or a special serial/netboot ramdisk), and each virtual console is in one separate screen thingy.
19:24 < svenl> and you can switch between them with screen hotkeys.
19:24 < bubulle_> h01ger: seems that the screen topic needs elaboration...and I didn't want to forget mentioning the localization issue
19:25 <+fjp> svenl: You'd need very good arguments to add it to initrd though.
19:25 < svenl> fjp: just add it to a new standalone netboot/serial ramdisk should cause no harm. none i can see at least.
19:25 < bubulle_> imho, all this is worth list discussion anyway
19:26 < waldi> yes
19:26 < bubulle_> I suggest we officially conclude the meeting now....which does not prevent discussions to go on
19:26 < svenl> indeed, i am fine to continue on list.
19:28 < bubulle_> as a side note, and sharing my personal feelings, I'd like to mention that I have been very happy to see svenl participate in this meeting. I sincerely hope this is the beginning of a new era after a difficult period and that personal difficulties being put aside are a good way to continue collaboration
19:28 < bubulle_> and if someone can restore the tocpi, please do so..:-)
19:29 -!- h01ger changed the topic of #debian-boot to: Next D-I team meeting July xxth xx:00UTC http://wiki.debian.org/DebianInstaller/Meetings | Debian installer development channel - http://www.debian.org/devel/debian-installer | If nobody answers, try debian-boot@lists.debian.org | status of today's builds: http://wiki.debian.org/DebianInstaller/Today
19:29 -!- waldi changed the topic of #debian-boot to: Debian installer development channel - http://www.debian.org/devel/debian-installer | If nobody answers, try debian-boot@lists.debian.org | status of today's builds: http://wiki.debian.org/DebianInstaller/Today


Attachment: signature.asc
Description: Digital signature

Reply to: