20:00:39 <bubulle> #startmeeting
20:00:56 <bubulle> #topic Alpha-1 release preparation
20:01:22 <bubulle> otavio: I don't think we have a release checklist as of now, do we?
20:01:44 <otavio> bubulle: no, we don't
20:01:49 <otavio> bubulle: well, sort of in wiki
20:01:56 <otavio> bubulle: but not  a complete one
20:02:07 <bubulle> #url http://wiki.debian.org/DebianInstaller/ReleaseProcess
20:02:16 <otavio> Well .. I think I should tell what is the current state
20:02:25 <otavio> We're doing somewhat well
20:02:36 <bubulle> #link http://wiki.debian.org/DebianInstaller/ReleaseProcess
20:02:45 <otavio> most of packages are not up to date (yes, I know that I need to get parted updated in d-i)
20:02:46 <bubulle> (sorry, that was for meetbot)
20:03:06 <otavio> bubulle: isn't it info?
20:03:32 <bubulle> what is needed for parted?
20:03:50 <otavio> but my current free time is splited between sleep, 5.0.2 update and d-i for squeeze
20:04:27 <otavio> bubulle: basically build partman-base and partitioner; you can do it if you want. I've built them locally with success (partitioner needs a porter for upload)
20:05:02 <otavio> kernel people has started .30 work and I believe we'll end up releasing with .30
20:05:21 <otavio> so would be nice if people could start checking their pet arches against .30 for we update it
20:05:34 <bubulle> oh, again another kernel version hunting party?
20:05:41 <wart> ouch
20:05:46 <otavio> i'll try to find a slot and look at kernel-wedge
20:05:52 <otavio> bubulle: well, not really
20:06:12 <otavio> bubulle: we're not near of release to us justify an delay on it
20:06:17 <wart> I don't think there will be lots of changes between .29 and .30.
20:06:32 <otavio> bubulle: waldi thinks we can have it done (ready for testing) in 30 days
20:06:38 <wart> (I mean, on d-i side)
20:06:52 <otavio> bubulle: near of what we need to test and fix bugs for release
20:06:58 <otavio> bubulle: so not that bad
20:07:01 <otavio> wart: me too
20:07:33 <bubulle> hmmm, 30 days brings us to mid-July and Debconf and we know that releasing during a Debconf is kinda dreaming
20:07:40 <otavio> today i need to upload lenny installer
20:08:15 <otavio> and then i'll start testing sid installer more carefully  to try to find bugs
20:08:23 <otavio> and would be nice if people does the same
20:08:36 <otavio> luk: netcfg patches, how are they going?
20:08:42 <bubulle> but it would be good if all partitioning things are in place, right?
20:09:01 <otavio> luk: do you think you can fix them in next days?
20:09:04 <otavio> bubulle: yes
20:09:45 <bubulle> otavio: so, I build partman-base? And we need a mips porter to build partitioner
20:10:08 <luk> I'm not sure I'll find the time, though it would be good to have the current patch tested...
20:10:39 <waldi> bubulle: what do you mean with "kernel version hunting party"?
20:10:52 <bubulle> waldi: running after kernels..:-)
20:11:56 <bubulle> otavio: correct me if I'm wrong but partman-base is not to be synced with parted, right?
20:12:03 <bubulle> but partitioner should?
20:12:27 <otavio> bubulle: sync?
20:12:42 <bubulle> built after parted enters unstable
20:12:53 <otavio> bubulle: yes, it needs to be
20:13:27 <bubulle> ah, but last parted actually *is* in unstable. Am I right?
20:13:55 <waldi> parted failed to build
20:14:37 <bubulle> waldi: on s390, right?
20:15:11 <waldi> yes
20:16:13 <bubulle> so, can we build partman-base now or is it safer to wait for parted to build on all arches
20:16:14 <mvz> question about alpha1.. I wonder..
20:16:22 <mvz> is much work wasted if we do an aggressive alpha1, followed by another alpha maybe a month later?
20:16:47 <bubulle> that was more or lessa my point and why I uploaded all packages
20:17:23 <bubulle> but releasing twice in one month has never been possible in all D-I history...:-)
20:18:25 <mvz> I'm lacking insight into whats involved in doing a release.. like how much of it is release "mechanics" and how much is careful testing, etc, which could perhaps be skipped over for an alpha?
20:18:26 <wart> Do we actually have Alpha1 goals?
20:18:36 <bubulle> indeed, either we release quite quickly with 2.6.29 kernels...or we wait for more than 1 month and release with 2.6.30,
20:18:59 <bubulle> #link http://wiki.debian.org/DebianInstaller/ReleaseProcess
20:19:04 <bubulle> mvz: ^^^
20:19:56 <mvz> bubulle: thx
20:20:02 <otavio> bubulle: it will depends on how well installer goes
20:20:14 <otavio> bubulle: in theory we could release with .29 without problem
20:20:15 <bubulle> on the packages side, we have nearly everything ready, except partman-base/partitioner/parted
20:20:15 <mvz> at least testing migration of udebs and CD/DVD build sound "expensive"
20:20:30 <waldi> bubulle: there is no 2.6.29 left
20:20:47 <otavio> mvz: not that much since currently is just "migrate all"
20:21:50 <bubulle> so, basically....*where* do we go? :-)
20:22:41 <bubulle> I see something like: sort out the part* things, then have work done on 2.6.30 kernels, maybe integrate netcfg udhcp patch and assemble the whoel thing
20:22:54 <mvz> wart: I think one goal is "put it all together" for the first time after lenny, to see what does/doesn't work
20:23:37 <bubulle> So, please let's take things one at a time and *decide* something
20:23:50 <bubulle> 1) parted
20:24:02 <bubulle> should we wait for it to build everywhere?
20:24:10 <bubulle> If so, who does monitor this?
20:24:16 * bubulle listens
20:24:20 <otavio> bubulle: it is built everywhere
20:24:25 <bubulle> s390?
20:24:28 <wart> which arches are missing?
20:25:04 <otavio> bubulle: and since libparted1.8-dev is built, partman-base will wait for it to be available if a arch is missing (alpha currentl y)
20:25:18 <otavio> https://buildd.debian.org/pkg.cgi?pkg=parted
20:25:54 <bubulle> ok, so no need to wait for parted, right?
20:26:01 <otavio> bubulle: yes, go ahead
20:26:13 <otavio> bubulle: after first arch goes in, it is safe to go
20:26:24 <bubulle> #agreed no need to wait for parted anymore
20:26:29 <bubulle> 2) partman-base
20:26:38 <otavio> bubulle: just because we had API/ABI changes
20:26:38 <bubulle> I build and upload. Any objection?
20:26:42 <otavio> no
20:26:43 <otavio> go
20:26:59 <otavio> bubulle: just be sure of doing it in a really updated chroot/system
20:27:04 <bubulle> #agreed bubulle builds and uploads partman-base
20:27:15 <bubulle> hey...I do build in clean chroots..:-)
20:27:35 <waldi> you do? :)
20:27:38 <bubulle> and I use pbuilder/cowdancer so they're up-to-date
20:27:58 <bubulle> waldi: much more since I build all my packages on my home server
20:28:12 <bubulle> 3) partitioner
20:28:19 <bubulle> who does it?
20:28:30 <otavio> tbm? aurel32?
20:29:34 <bubulle> ring ring knock knock
20:30:17 <bubulle> #info We need to find someone to build partitioner on mips
20:30:30 <bubulle> 4) other remaining packages to build
20:30:48 <aurel32> otavio: yes?
20:30:55 <waldi> bubulle: isn't there a mips porter box?
20:31:36 <bubulle> waldi: I don't feel comfortable enough for this. I prefer leaving porter builds to other ppl..:-)
20:31:48 <aurel32> what is really needed?
20:32:03 <bubulle> aurel32: build partitioner from D-I SVN on mips
20:32:24 <aurel32> I can do it yes
20:33:27 <bubulle> aurel32: see installer/doc/devel/package-upload.txt for the few tricks and checklist to build D-I udebs from SVN
20:33:48 <bubulle> #agreed aurel32 will build and upload partitioner
20:34:06 <otavio> cool
20:34:14 <bubulle> otavio: I'm trying to look around your long list to see what's still needed
20:34:27 <bubulle> - zipl-installer
20:34:30 <bubulle> (s390)
20:34:47 <otavio> For Alpha1, we're more or less doing fine. Mostly we need packages uploaded and tests
20:35:17 <otavio> After parted is done I'll upload installer to get it built and check for any remaining problem
20:35:29 <otavio> it won't be a release one, but to get it done only
20:36:17 <bubulle> aboot-installer, efi-reader, elilo-installer, partman-palo, partman-prep, s390-dasd, s390-netdevice, vmelilo-installer
20:36:33 <bubulle> (all those only have translation changes, so we can just skip them)
20:37:01 <wart> Is alpha still a release arch, btw?
20:37:11 <bubulle> #agreed after parted/partman-base/partitioner are done, otavio builds and uploadds debian-installer
20:38:03 <bubulle> otavio: we need to decide about luk's patch to netcfg
20:38:18 <fezie> wart: afaik yes
20:38:22 <otavio> bubulle: I'd like to push it but I also don't like to drop previous support right now
20:38:32 <otavio> bubulle: or luk finds the time to fix it or someone does it
20:38:39 <bubulle> otavio: the patch is easy to revert..:-)
20:38:43 <otavio> bubulle: or  I do, but currently no ETA for it
20:38:54 <otavio> bubulle: well, it is not really the case for reverting it
20:39:07 <otavio> bubulle: but to keep other support around while we're testing it
20:39:26 <otavio> bubulle: so it is just the case of a package removal to get it tested
20:40:05 <otavio> bubulle: and also, it makes eaiser to check  the changes individually
20:40:27 <otavio> bubulle: I'm not a big fan of " drop something and add something new in same turn "
20:40:52 <otavio> bubulle: those are two really different things so two patches looks sanner for me
20:41:17 <bubulle> OK, so you suggest luk keeps the code that uses dhclient from dhcp3-udeb and only adds calls to udhcp if the dhclient is not found?
20:41:56 <bubulle> then we upload netcfg and we stop including dhcp3-udeb in pkg-lists?
20:41:57 <otavio> bubulle: yes; as I suggested on the patch review
20:42:04 <otavio> bubulle: exactly
20:42:06 <luk> it's rewriting anyway as it does not work the way pump and dhcp3 work btw
20:42:31 <otavio> luk: sure but you'll work in a specific swtich case
20:42:40 <luk> nope
20:43:07 <luk> you put everything there is now in an if or else statement and put the rest for udhcp
20:43:26 <luk> as they are really not at all compatible in how you give the options
20:44:07 <otavio> I'll need to look at the patch again to be able to discuss it
20:44:17 <otavio> luk: do you mind to reply to the mail I sent?
20:44:25 <otavio> luk: it makes all easier for both
20:44:53 <bubulle> so, can we summarize this to "discussion continues about udhcp patch"?
20:45:14 <otavio> bubulle: we want to have it in
20:45:21 <otavio> bubulle: we just need to decide how
20:45:57 <bubulle> and when? Having it for alpha1 would be good but it shouldn't delay alpha1 if some more work is needed?
20:46:24 <otavio> yes
20:46:35 <otavio> also because it won't make a huge difference right now
20:46:50 <bubulle> luk: do you agree with that approach?
20:47:16 <luk> sure, I'm working on the reply :-)
20:47:23 <otavio> luk: cool and thx
20:47:28 <CIA-2> debian-installer: 03aurel32 * r58983 10packages/partitioner/debian/changelog: releasing version 0.45
20:47:35 <CIA-2> debian-installer: 03aurel32 07packages * r58984 10/partitioner/0.45/: tagging version 0.45
20:48:08 <bubulle> #agreed more discussion about netcfg patch and the way to make the switch-->mailing list
20:48:30 <bubulle> and thanks to aurel32, Lucky Luke of package building
20:49:11 * bubulle scratches head to find if there are more things to discuss wrt alpha1 release
20:49:32 * otavio will be right back
20:49:33 <bubulle> (btw, I called it mentally "alpha1". otavio, is this the say you see it?)
20:49:58 <bubulle> ah, we lost our release manager...:(
20:50:11 <otavio> back
20:50:20 <otavio> yes; i agree
20:50:34 <otavio> we shouldn't worry too much about wide testing
20:50:46 <otavio> just release it and deal with bugs later
20:51:00 <otavio> obviously we won't release a know broken image heh
20:51:39 * youpi back
20:52:41 <otavio> are we all in same page regarting it?
20:52:46 <bubulle> otavio: do you think it will be possible to build a rough schedule for the release?
20:53:07 <otavio> bubulle: yes; but after first tests of parted only
20:53:22 <otavio> bubulle: since it can severely break the usability
20:54:01 <bubulle> ok; then
20:54:35 <bubulle> I listed something called "Fill in StatusUpdate" but I don't find any such wiki page..:-)
20:55:14 <bubulle> anyone wanting to say things about release goals or work in progress?
20:55:38 <bubulle> youpi: for c-s for instance..:-)
20:55:53 <youpi> well, we're still waiting for Franz' mail :)
20:56:52 <bubulle> yes. I personnally didn't do much more testing since I prepare test images
20:57:01 <youpi> I haven't either
20:57:24 <bubulle> the first feedback was not that bad, but I somewhat "fear" Frans mail..:-)
20:57:55 <youpi> same for me :)
20:58:04 <bubulle> please note that last localechooser upload will trigger c-s install on the target system for all languages that were using kbd previously
20:58:15 <youpi> ok
20:58:32 <youpi> I guess there's no better way to have it tested :)
20:58:35 <bubulle> so we will have reports of questions for keyboard being re-asked, I think
20:58:56 <bubulle> that should happen until we replace kbd-config by c-s-udeb
20:59:19 <otavio> bubulle: yes; that ought to be add on current know issues
20:59:30 <otavio> bubulle: to allow us to point people there and avoid useless questions
20:59:46 <otavio> bubulle: obviously if we release with this regression
20:59:55 <bubulle> otavio: note that I haven't checked this yet. I suspect that it will happen, that's all..:-)
21:00:58 <otavio> bubulle: ok then
21:01:08 <CIA-2> debian-installer: 03bubulle * r58985 10packages/partman/partman-base/debian/ (changelog compat control):
21:01:08 <CIA-2> debian-installer: Bump debhelper compatibility to 6
21:01:08 <CIA-2> debian-installer: Add myself to uploaders
21:01:13 <bubulle> I have not many more things to add to the meeting, indeed
21:01:25 <bubulle> and I'd like to go to sleep early..:-)
21:01:26 <otavio> I think that all possible issues are discussed for alpha1
21:01:56 <otavio> Just before you go to sleep;
21:02:36 <otavio> I'd like to ask how people feel about current process that we're using in d-i since our first meeting. The changes how the RM has been working and like
21:03:45 * cjwatson shows up a bit late
21:03:48 <cjwatson> *cough8
21:04:11 <bubulle> well, /me would be reliefed to have less things to push all around and sometimes feel like running as a headless chicken during meetings..:-)
21:04:54 <bubulle> sometimes with the feeling that this is needed to keep things alive...
21:05:30 <otavio> right
21:05:52 <otavio> i've been trying to push things as I can
21:05:55 <cjwatson> if we're in AOB time, has anyone had any clear word on whether the UUID changes seem to be working out OK?
21:06:00 <cjwatson> those were my main goal for alpha1
21:06:00 <otavio> and I think it has been working much better
21:06:17 <cjwatson> I suspect they will only work *really* well for people using grub2
21:06:38 <otavio> cjwatson: I think it is but we're stuck due parted arrivel on sid
21:06:51 <fezie> cjwatson: they get anyway only used with grub2
21:06:56 <otavio> cjwatson: but it is going to be solved today I guess
21:06:57 <CIA-2> debian-installer: 03bubulle * r58986 10packages/partman/partman-base/debian/changelog: releasing version 130
21:06:58 <fezie> and grub2 in lenny already uses them
21:07:08 <CIA-2> debian-installer: 03bubulle 07packages * r58987 10/partman/partman-base/130/: tagging version 130
21:07:10 <fezie> just not the grub-installer generated file did
21:07:18 <otavio> cjwatson: what is the missing points for us to be able to enable ext4?
21:07:25 <otavio> cjwatson: parted should be in shape
21:07:27 <cjwatson> fezie: UUIDs were also relevant for partition mounting; they aren't just a boot loader thing
21:07:36 <fezie> ah
21:07:39 <cjwatson> otavio: hmm, I think all the blockers were linked from the wiki
21:07:47 * cjwatson looks
21:08:41 <cjwatson> otavio: nothing, as far as I can see, assuming proper use of new parted
21:09:00 <bubulle> otavio: I propose stopping the formal meeting now..unless you guys want to continue with things "recorded"
21:09:00 <cjwatson> otavio: shall I turn it on and we'll see what happens?
21:09:02 <mvz> one open question was about non-grub2 bootloaders, iirc
21:09:19 <cjwatson> mvz: wrt UUIDs you mean?
21:09:24 <otavio> cjwatson: I guess so
21:09:30 <mvz> cjwatson: no, ext4
21:09:31 <otavio> cjwatson: better now then later
21:09:34 <cjwatson> oh, right
21:09:39 <otavio> bubulle: yes, please go ahead
21:09:42 <otavio> bubulle: for me it is ok
21:09:43 <cjwatson> well, they can always have /lib/partman/check.d scripts added
21:09:53 <cjwatson> but if it's not the default it doesn't seem critical to add those
21:10:17 <bubulle> otavio: partman-base is uploaded (and built with the right libparted1.8-dev)
21:10:26 <bubulle> #endmeeting