Log from the Debian Installer team meeting of November 29th 2008, 20:00UTC
--------------------------------------------------------------------------

People who speak below:

otavio    : Otavio Salvador
anibal    : Anibal Monsalve Salazar
fjp       : Frans Pop
bubulle   : Christian Perrier
Ryan52    : Ryan Niebur
waldi     : Bastian Blank
nyu       : Robert Millan
faw       : Felipe Augusto van de Wiel
Ganneff   : Joerg Jaspert
p2-mate   : Peter de Schrijver
taggart   : Matt Taggart
ths       : Thiemo Seufer

People whose nicknames are mentioned during the meeting but not present:

franklin  : Franklin Piat
luk       : Luk Claes

Hours are European Time (UTC +0100):

21:00 -!- bubulle changed the topic of #debian-boot to: D-I meeting in progress : http://wiki.debian.org/DebianInstaller/Meetings/Coordination
21:00 < otavio> Ganneff: no; we did it in propouse :P
21:01 < bubulle> So, mostly the agenda is about releasing RC1....
21:01 < bubulle> RC2
21:01 < otavio> right
21:01 < bubulle> kernel udeb updates
21:01 < anibal> are we using meetbot.debian.net
21:01 < anibal> ?
21:01 < bubulle> otavio: you uploaded today
21:01 < otavio> This is my current talk. I guess we'll be ok with it until Monday.
21:01 < bubulle> anibal: if you want
21:01 < otavio> anibal: start it
21:02 < otavio> anibal: and we start it once again
21:02 < bubulle> h0lger told me it is listening here
21:02 < bubulle> otavio: what remains to be done?
21:02 < anibal> bubulle, should start it, he's charing this meeting :)
21:02 < bubulle> sure, but I'm too dumb to know how to use it
21:02 < otavio> bubulle: few arches are not yet uploaded 
21:02 < anibal> ok
21:03 < bubulle> Oh? i forgot: who's around?
21:03 < otavio> bubulle: so tests
21:03 < otavio> bubulle: and then close the bugs about sparc and like
21:03  * luk_extremadura_ is around
21:03  * ths is lurking
21:03 < bubulle> otavio: with today's uploads, the dailies will have the new kernels, right?
21:04 < otavio> bubulle: yes
21:04 < bubulle> when tests will show things are OK with kernels, what has to be done after?
21:04 < otavio> bubulle: i can try to check for the missing arches later today.
21:05 < otavio> bubulle: basically update release announce and close the bugs. We'll then migrate it all when starting RC2 process
21:05  * otavio hopes in few days ... or so
21:05 < bubulle> ok...so basically, for kernel updates, things are on their way
21:05 < otavio> yes
21:05 < bubulle> I propose we move to other points on the agenda one by one
21:05 < bubulle> "Fixed issues. Actions needed for them:"
21:06 < bubulle> "install guide release"
21:06 < bubulle> I think that nothing specific is needed for that. fjp, if around can you confirm?
21:07 < otavio> We need to know when Frans is planning to do the release of it (if he plans to do another one)
21:07 < otavio> But we are OK on this AFAIK
21:07 < bubulle> otavio: I don't think he plans to do another one
21:07 < bubulle> that item was on the RC2 todo list but was completed very quickly
21:07 < bubulle> so we can probably safely assume that things are OK, here
21:07 < otavio> bubulle: yes
21:08 < bubulle> Frans will comment if needed
21:08 < bubulle> "add USB storage modules to Sparc images to allow loading firmware"
21:08 < bubulle> iirc, done by otavio and no need for any specific action
21:08 < otavio> those looks OK in dailies and will be part of RC2 (same as kernel updates)
21:08 < otavio> yep
21:08 < bubulle> ok, next:
21:08 < bubulle> "Add uinput module for better support of some Braille devices"
21:09 < otavio> same
21:09 < bubulle> ok
21:09 < otavio> We need testing on this one.
21:09 < otavio> I've not seen any ack about it get done right. It would be nice to ping Samuel and check it
21:09 < bubulle> that's part of the tests that have to be done when dailies will have the new kernels, right?
21:10 < bubulle> any volunteer to ping Samuel after the meeting?
21:10 < otavio> IIRC it has been present in current udebs
21:10 < bubulle> one more reason to test now, then..:-)
21:11 < otavio> exactly ... it was done in kernel-wedge upload
21:11 < otavio> in 2.53
21:11 < otavio> 18 of november
21:11 < bubulle> ok, no volunteer, let's hope that Samuel will read this, then
21:11 < bubulle> "allow user to return to syslinux VESA menu from help screens"
21:11 < bubulle> work was done by franklin and fjp
21:12 < otavio> It looks OK on current dailies. Frans and Franklin has done a bunch of work to get it done.
21:12 < bubulle> so, it's considered done and only needing a d-i upload (of course)
21:12 < otavio> yes
21:12 < otavio> on the way for RC2
21:12 < bubulle> ..... we're done with the easy stuf..:-)
21:13 < bubulle> Now for the nasty things.
21:13 < otavio> hehe
21:13  * bubulle takes some Xibonquinha (kindly provided by faw)
21:13 < otavio> Xibonquinha?
21:14 < bubulle> s/n//
21:14 < bubulle> "#505111: rescue-mode md auto-assembly could cause data loss"
21:14 < bubulle> I tried to "wake up" the issue with my stupid suggestions
21:15 < faw> actually is Xiboquinha :) without 'n'
21:15 < otavio> Yes; it looks stuck 
21:16 < otavio> I'm unsure if we should consider it RC; maybe just note it in errata and let it for lenny+1
21:16 < bubulle> my rough feeling is that it can be ignored for lenny (as fjp mentioned, it requires invasive changes) and documented in the errata
21:16 < otavio> ack
21:16 < Ganneff> foodtime, mr. broken keyboard
21:16 < bubulle> luk_extremadura_: we might consider tagging "lenny-ignore"
21:17 < otavio> I didn't see this comment from Frans. But I fully agree on that.
21:17 < bubulle> #505111
21:17 < otavio> luk_extremadura_: please comment on that once you can.
21:17 < bubulle> otavio: Frans did not suggest to ignore it but this is the conclusion I drew from his comments....
21:17 < bubulle> otavio: we're moving to the restaurant. I ping you when there
21:18 < otavio> OK; 
21:18  * otavio holds 

<skip unrelated discussion about Brazilian alcohols and how nice was Debconf 4>


21:36 < bubulle> ok, I'm here
21:36 < bubulle> soooooo
21:37 < otavio> :-)
21:37 < bubulle> we were talking with luk_extremadura_ while coming around and he proposes that the rescue issue is downgraded
21:37 < otavio> bubulle: we were talking about the good things of Brazil, specially caipirinha ;-)
21:38 < bubulle> I think it's better to keep it RC, document in the errata and tag lenny-ignore
21:38 < otavio> bubulle: yes; I think I agree with you on that
21:38 < bubulle> and luk_extremadura_ is OK with that
21:38 < otavio> good
21:38 < otavio> luk_extremadura_: thx
21:38 < bubulle> next unfixed issue: "Build failures on Sparc"
21:39 < bubulle> actually,this one is just "need to be checked"
21:39 < otavio> bubulle: should be fixed with today's upload
21:39 < otavio> yes
21:39 < bubulle> yes, it rejoins the kernel stuff
21:40 < bubulle> let's move on...to some others that maybe need more discussion:
21:40 < bubulle> #504721: Console broken on debian-installer on Sparc
21:41 < otavio> I personally didn't follow it. It looks from Jeremy comments that it is a kernel issue but we've not yet get a confirmation about it
21:41 < bubulle> hmm, no, there is a proposed fix by Jurij
21:41 < otavio> humm; let me see if I can find it
21:42 < bubulle> luk_extremadura_ mentioned he could be in position to do some testing
21:43 < bubulle> also, Jurij told that passing "console=ttyS0" at the boot prompt is a way to work this around
21:44 < bubulle> so, maybe we could try Jurij proposed fix 
21:44 < otavio> oh; found it.
21:44 < otavio> He said he was going to check about reopen-console but he has not yet replied to it
21:45 < otavio> We could ping him to see if he has got any progress on that
21:45 < bubulle> it needs someone to patch rootskel, build it on sparc and rebuild D-I for Sparc with it
21:45 < bubulle> Jurij is not online right now, apparently
21:46 < otavio> and check if it doesn't introduce another issues; to avoid more regressons on sparc
21:50 < bubulle> luk_extremadura_ would be OK to test
21:50 < bubulle> he needs some detail about what needs to be done:
21:50 < bubulle> - build rootskel with Jurij patch
21:51 < bubulle> - rebuild d-i with the new udeb
21:51 < bubulle> - test
21:51 < otavio> I can do it with him tomorrow or on Monday; I have a dinner after meeting and if I do it tomorrow during the day my gf would kill me :P
21:52 < otavio> Otherwise someone might guide him on that before I'm available.
21:52 < bubulle> otavio: well, luk_extremadura_ will be travelling tomorrow
21:52 < bubulle> he can probably test on Monday
21:52 < otavio> bubulle: so on Monday
21:52 < otavio> good
21:52 < bubulle> the best would be someone building a D-I image for Sparc but I'm unsure if someone around can
21:54 < bubulle> move to next "unfixed issue"?
21:54 < bubulle> (sorry for the jerky communication here...we're having dinner at the restaurant: just picture the thing!)
21:54 < bubulle> "#504760: s390 installer doesn't work on 3215 console running as a VM guest"
21:55 < otavio> I have not followed it. Someone did?
21:55 < bubulle> I exhumated that one today....:-|
21:55 < bubulle> otavio: fjp commented on it
21:57 < bubulle> if no-one digs into it (waldi or fjp, maybe?) then the issue would be as earlier: document it, tag lenny-ignore and move on
21:58 < otavio> yep; the user has not provided the asked info
21:58 < otavio> waldi: could you take a look on it?
21:59 < bubulle> Then, we have non RC issues:
21:59 < bubulle> "Fix missing dm-crypt in rescue mode"
21:59 < otavio> Is it just the module that is required?
22:00 < bubulle> #506068
22:00 < bubulle> yes, the module is just missing
22:00 < otavio> indeed; it looks to be a missing module
22:01 < otavio> I can look into it; if required I do another massupload
22:01 < otavio> Let it to me; Can you update LennyPrepRC2?
22:01 < CIA-1> debian-installer: waldi * r56808 packages/rootskel/debian/changelog: debian/changelog: Add bug closure.
22:02 < waldi> rootskel needs an upload
22:02 < CIA-1> debian-installer: otavio * r56809 packages/kernel/linux-kernel-di-arm-2.6/debian/changelog: releasing version 1.35
22:02 < bubulle> waldi: ah, now I remember...while looking at this issue I noticed your change
22:03 < bubulle> So, the upload would fix#504760? 
22:03 < CIA-1> debian-installer: otavio packages * r56810 /kernel/linux-kernel-di-arm-2.6/1.35/: tagging version 1.35
22:03 < CIA-1> debian-installer: otavio * r56811 packages/kernel/linux-kernel-di-armel-2.6/debian/changelog: releasing version 1.30
22:03 < CIA-1> debian-installer: waldi * r56812 packages/rootskel/debian/changelog: debian/changelog: Prepare to release (1.70).
22:03 < CIA-1> debian-installer: otavio packages * r56813 /kernel/linux-kernel-di-armel-2.6/1.30/: tagging version 1.30
22:04 < CIA-1> debian-installer: otavio * r56814 installer/doc/devel/kerntabl: Update kernel version to 2.6.26-11
22:05 < CIA-1> debian-installer: otavio * r56815 packages/kernel/massbuild.versions: Update kernel version to 2.6.26-11
22:05 < bubulle> OK, so #504760 would be fixed by a rootskel upload
22:05 < bubulle> "Fallback languages not working when not English"
22:05 < otavio> There's a patch sent today about it
22:05 < otavio> bubulle: could you try the patch?
22:06 < bubulle> otavio: but not addressing the right bug, according to fjp
22:06 < bubulle> and I confirm he's very probably right
22:06 < otavio> bubulle: the today patch? 
22:06 < bubulle> very unfortunately, my local builds just f..... fail to boot in my virtual machine here
22:06 < bubulle> yes, the today patch
22:06 < bubulle> I mail Jérémy about this
22:07 < otavio> I'd be interested in the bug number about that issue. Could you add them to prep wiki page?
22:07 < CIA-1> debian-installer: waldi packages * r56816 /rootskel/1.70/: Release rootskel (1.70).
22:07  * otavio say's kernel update is complete
22:07 < bubulle> #502244. It is already there..:-)
22:07 < bubulle> I just didn't copy it in the meeting agenda
22:08 < otavio> bubulle: frans said that there's another related issue but I fail to remember which one
22:08 < bubulle> that one..:-)
22:08 < otavio> bubulle: ah ok great
22:08  * anibal sent the email to Samuel Thibault
22:09 < otavio> anibal: thx
22:09 < bubulle> #502240 and #502244 are the bugs related to these cdebconf issues
22:09 < bubulle> #502240 has a patch, #502244 hasn't
22:10 < otavio> bubulle: I added the other one in prep wiki page
22:11 < otavio> next ?
22:11 < bubulle> let me finish my soup..:-)
22:11 < bubulle> "Display problems in Dzongkha installs"
22:12 < otavio> Did somone looked at it?
22:12 < otavio> I see no move on this one
22:12 < bubulle> no move
22:13 < bubulle> david-e would be qualified, maybe
22:13 < bubulle> that seems *very* hairy
22:13 < bubulle> and certainly not a blocker
22:13 < bubulle> I propose documenting it and blah
22:13 < bubulle> #505408
22:14 < otavio> ack
22:14 < otavio> You could start working on the errata for this since it is present in RC1 too
22:14 < bubulle> sure
22:14 < otavio> And then we have one less thing to do later ;-)
22:14 < bubulle> Next: "Add the LXDE tsk to the XFCE CD"
22:15 < otavio> It is kind of sorted out with the mixed CD
22:15 < bubulle> fjp proposed an image to test
22:15 < bubulle> otavio: yes
22:15 < otavio> Frans did a awesome job for this issue and it looks a great improvement
22:16 < bubulle> I think I saw some feedback which was positive
22:17 < bubulle> from what I saw, this is worth being activated (on debian-cd, I think?)
22:17 < bubulle> Sledge: ^^^^^^^
22:17 < bubulle> Next issue also has been well worked out: " 
22:17 < bubulle> Multiarch CD directly boots in the amd64 installer"
22:18 < bubulle> franklin_: around?
22:18 < otavio> This is a very small issue and I believe it's not worth the risk of a syslinux new upstream release only for it. I'd prefer to add it to errata.
22:19 < otavio> Other problems were workarounded by Frans and Franklin
22:19 < bubulle> but does the fix really need a new syslinux?
22:20 < bubulle> ok, so if the main problem is sorted with commited changes to the build system, then I think it's OK
22:20 < otavio> IIRC the only remaining issue is that if he is on help and press enter it boots the wrong kernel (i386 iirc)
22:20 < otavio> fjp: franklin_: can you confirm the above later?
22:21 < bubulle> so, we just document the remaining bit in the errata, then
22:21 < otavio> bubulle: if it is the last remaining issue; yes.
22:22 < bubulle> Last issue to talk about: "grub-installer: grub2 config for Windows partition does not boot"
22:22 < bubulle> no idea about this..:-)
22:22 < otavio> about this one I'd like to get a feedback from nyu
22:23  * otavio asked him to join here
22:23 < otavio> nyu: thanks
22:23 < otavio> nyu: [19:22] < bubulle> Last issue to talk about: "grub-installer: grub2 config for Windows partition does not boot"
22:23 < nyu> IIRC I sent a patch for that one
22:23 < bubulle> #473401
22:23 < otavio> nyu: we're finishing RC2 meeting and this is one thing I'd like a guidance from you
22:24 < bubulle> as I read, osamu commented after nyu's patch
22:24 < nyu> sure.  what would you like to know?
22:25 < anibal> nyu, what's your name (for the minutes)?
22:25 < nyu> I find osamu's report a bit confusing.  but I already commented on it, I think it's in #502446
22:25 < nyu> anibal: Robert Millan
22:25 < anibal> thanks
22:26 < nyu> brb
22:26 < bubulle> nyu: do you consider this must be fixed for D-I RC2 (and therefore lenny) or we can live with it, if documented in errata
22:27 < otavio> nyu: I saw few comments from Frans; did you test the points that we giving against your patch? 
22:29 < otavio> nyu: i did't get what osamu said about translation. Last patch from him has no po changes
22:29 < bubulle> I would be somewhat reluctant to have this now in D-I....isn't that taking some risk of more breakages than fixes?
22:30 < otavio> bubulle: somewhat. grub-probe runs if user installs directly
22:30 < otavio> bubulle: so if we have a bug here it is in grub-probe and grub-probe is shared in grub and grub2
22:30 < otavio> bubulle: I somewhat think it is safe to use it
22:31  * otavio waits for nyu comment about the above
22:32 < otavio> nyu: ?
22:32 < otavio> oh he is away
22:32 < bubulle> I'd really prefer going the conservative way, here
22:33 < bubulle> that patch should have been tested back in May when it was proposed
22:33 < otavio> yes; agree on that
22:33 < bubulle> so, again, document and move one...:-|
22:34 < otavio> i believe it is much risk to fix it changing the convert() then getting rid of it (moving to grub-probe)
22:34  * bubulle is not sure to understand (late+restaurant+wine+no_sleep) but takes otavio's word on it
22:35 < bubulle> we reached the final point to discuss, mostly
22:35 < otavio> bubulle: the logic behind convert is hard to follow is many ways. grub-probe is what grub has been using for same thing
22:36 < bubulle> the next (opened) point is: what road to follow to release?
22:36 < otavio> bubulle: you're going to prepare a draft of new errata items?
22:36 < nyu> re
22:36 < otavio> nyu: comment on the above, plz
22:36 < bubulle> otavio: sure but that will be slow (trvelling tomorrow, then busy week)
22:37 < otavio> bubulle: right; no problem. I'll check dm-crypt thing tomorrow and see if I can make a image for luk on monday
22:37 < nyu> otavio: which comments do you mean specifically?
22:38 < otavio> nyu: about grub-probe being used by grub and grub2 (am I right about it?) and how difficult would be fix current convert function instead of getting rid of it now?
22:38 < nyu> otavio: yes
22:39 < nyu> let me check about convert
22:39 < bubulle> otavio: maybe we could schedule a "release meeting" by the end of the week?
22:39 < otavio> bubulle: it is ok for me
22:40 < nyu> what convert() is doing is simply duplicate what grub-mkdevicemap does
22:40 < otavio> nyu: yes; i know
22:40 < nyu> grub-mkdevicemap generates device.map, and grub-probe uses that
22:40 < otavio> nyu: however we'd prefer to not change it so late in release
22:40 < otavio> nyu: so would be hard to fix it?
22:40 < nyu> I think it should have been changed earlier, yes
22:40 < otavio> nyu: could you take a look in convert for us?
22:41 < anibal> bubulle, Satday at 20:00 UTC?
22:41 < otavio> nyu: if it turns to be hard and/or risky, we document it and let it for next release
22:41 < nyu> I think it's safer to use grub-probe at this point.  it's not like not calling grub-probe is going to avoid any problem that may exist
22:42 < otavio> anibal: it works for me
22:42 < nyu> since grub-probe is being used anyway.  and it's known to work well now
22:42 < bubulle> anibal: I don't know.
22:42 < otavio> nyu: can you take a look on the bug and comment on the points that Frans has expressed on it?
22:43 < nyu> I think I did?
22:43 < nyu> let me check
22:43 < otavio> nyu: so we decide it later, in ml.
22:43 < otavio> nyu: no
22:43 < otavio> nyu: there're few points still open
22:43 < otavio> nyu: and that need testing
22:43 < bubulle> anibal: taking yet another saturday night is high-risk for me at home!
22:43 < nyu> otavio: which ones?
22:43 < otavio> anibal: it could be at friday
22:43  * bubulle would rathe rgo for a meeting on sat afternoon
22:43 < otavio> anibal: so support and like
22:43 < otavio> ops
22:43 < otavio> nyu:   ^
22:44 < otavio> bubulle: it works for me too; np
22:44  * otavio really needs to go now
22:44 < nyu> otavio: are you referring to http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=473401#23 ?
22:44 < otavio> nyu: yep
22:44 < bubulle> anibal: maybe try to hget a moment that's convenient for fjp also (apparently tonight wasn't)
22:44 < anibal> ok
22:45 < fjp> It's not about convenient.
22:45  * bubulle doesn't really go, but shuts the laptop off to enjoy the dessert, at least
22:45 < nyu> otavio: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=473401#41
22:46 < otavio> humm
22:46 < nyu> otavio: this was from June.  IIRC we didn't get any new report about this
22:47 < nyu> (regarding point 2)
22:47 < nyu> point 3 is no problem, and point 1 is not really something that we can ever fix (except by ditching the whole idea that BIOS drives can be guessed)
22:48 < otavio> About point 2: 
22:48 < otavio>  multipath has been added?
22:48 < otavio> or better: could you mail ml about what is still open in point 2?
22:48 < otavio> mail to the bug
22:50 < otavio> I'm uncertain if we should or not fix this for lenny. grub-probe looks very stable now but I am still uncertain. Please make clear to us what is still untested and needs checking so we can decide with that information
22:50  * otavio now needs to go
22:50 < nyu> I don't know the status of multipath, except that GRUB never supported it.  I know grub-installer has some hack which is supposed to make it work, but I have no idea if my proposed change would break that or not
22:51 < nyu> the other stuff mentioned in my mail (IDA, I2O, etc) has been supported for a while now
22:51 < otavio> nyu: ping agx to get it; He can help on that.
22:51 < nyu> plus virtio, which IIRC wasn't supported before that
22:51 < otavio> nyu: please; make this update in the bug. So we can use it to point the right people to it
22:51 < otavio> agx for example.
22:52  * otavio is out now
22:52 < otavio> bye
22:52 < nyu> will do
22:57  * bubulle shuts down the laptop (low battery and no AC)