Log from the Debian Installer team meeting of June 15th, 2009, 20:00UTC
-----------------------------------------------------------------------

People who speak below:

 bubulle   : Christian Perrier
 dilinger  : Andres Solomon
 elmig     : Miguel Figuereido
 maks      : Maximilan Attems
 otavio    : Otavio Salvador
 panthera  : Daniel Baumann
 pusling   : Sune Vuorela
 slackydeb : Luca Favatella
 trave11er : Jurij Smakov
 waldi     : Bastian Blank

22:00 < bubulle> OK; time to begin the meeting...with 3 people around
22:00  * bubulle guess this will be a short one
22:01 < bubulle> otavio: so, let's talk about releas preparation...
22:01 < otavio> slackydeb: there?
22:01 < otavio> bubulle: ok
22:01 < otavio> Well
22:02 < otavio> we had a quite unluck time since 2.6.29 didn't migrate to testing and then we'll need to move to 2.6.30 for release since we'll need to wait it to be ready to migrate 
22:02 < otavio> or testers will need to install sid
22:02 < bubulle> ok
22:02 < otavio> luk_: there?
22:02 < otavio> oops
22:02 < otavio> bubulle: didn't see you have hilight him 
22:03 < bubulle> is 2.6.30 something that we have good chances to enter in testing?
22:03 < otavio> I'll look at updating the kernel-wedge package against it
22:03 < otavio> and ask porters to upload the udeb packages
22:03 < otavio> we also have a ftbfs in s390 in parted
22:04 < otavio> i asked for few packages to be installed on the developer machine to be able to compile the package there
22:04 < otavio> will look at it tomorrow, probably
22:04 < bubulle> yes, so parted and kernel are the blockers right now?
22:04 < otavio> yes
22:04 < slackydeb> otavio: here
22:05 < bubulle> and otavio builds it manually
22:05 < bubulle> any kernel ppl around?
22:05 < bubulle> waldi: ?
22:05 < bubulle> would be good to know about plans for puching 2.6.30 to testing and/or upload 2.6.31 (if such beast already exists)
22:05 < panthera> .oO(i'm here, basically listening only i guess)
22:06 < otavio> slackydeb: please be around since i'd like to talk to you about kfreebsd 
22:06 < otavio> bubulle: from what i talked yestarday with kernel people (maks) it looks like they're preparing a upload soon
22:06 < slackydeb> otavio: ok. In the meantime I'm testing a patch of netcfg to propose to debian-boot list as soon as possible :)
22:06 < otavio> i guess it can be good enough to get it into testing
22:06 < bubulle> otavio: we can keep some time for kfreebsd stuff after the release preparation part...which won't take long
22:06 < waldi> oh, here
22:07 < otavio> waldi: please tell us the kernel team plans regarting the testing migration and all
22:08 < waldi> it is up to the release team. it will not go everywhere without some larger amount of force or removals
22:08 < maks> also here
22:08 < waldi> also we need 2.6.30-2
22:09 < bubulle> hmm, so even if we solve out the parted problem, we'll be stuck by kernel stuff, right?
22:09 < otavio> waldi: what kind of removals?
22:09 < panthera> waldi: 2.6.30-2 as in debian version or abiname?
22:09 < waldi> otavio: modules packages
22:09 < waldi> panthera: version
22:09 < maks> diverse flavours and stupid modules
22:10 < otavio> panthera: any plan to get modules updated?
22:10 < panthera> otavio: i should get arround to it for a first upload soon (tomorrow or on wed)
22:10 < waldi> otavio: will get rejected most likely, see the gcc-ming* mail
22:11 < otavio> waldi: mind to give me a pointer to the thread?
22:12 < otavio> waldi: so i could look at it
22:12 < bubulle> so, from what I understand, we need the release team to put some priority on having 2.6.30 in testing so that we can release our first alpha
22:12 < otavio> bubulle: i guess we can get RM support for it
22:12 < waldi> otavio: <20090705215258.GA10868@thorin> on -release
22:12 < bubulle> luk_: ^^^^ would be good if you could put some priority in this
22:13 < bubulle> I think that anyway, having a new kernel in testing 5 months after the lenny release is really needed 
22:13 < trave11er> luk is on vacation this week
22:13 < otavio> waldi: what it has to do with modules and testing migration?
22:14 < waldi> otavio: gpl+source
22:14 < otavio> waldi: sorry; i didn't get it
22:14 < waldi> otavio: since when is REJECT a testing problem?
22:15 < otavio> waldi: we're wondering about kernel and modules. For us the gcc-ming* doesn't matter (from d-i pov)
22:15 < otavio> waldi: or am I missing anything?
22:15 < waldi> otavio: i spook about REJECT, the mail lists a REJECT reason
22:16 < bubulle> ok, anything more to say about kernel-related issues
22:16 < bubulle> this time, I really think we need everybody to work towards having a new kernel in testing. 
22:16 < waldi> no
22:17 < otavio> bubulle: yes; we need the kernel and release team support on that
22:17 < bubulle> no, what? no, we can't work towards having a new kernel in testing?
22:17 < otavio> heh
22:17 < waldi> no for more
22:17 < otavio> waldi: when is -2 planned?
22:18 < otavio> waldi: and when the last -1 ABI upload is planned?
22:18 < bubulle> could there be a pause somewhere in new upstream for kernel? New upstream could then go to experimental by the time a new kernel makes it in testing....
22:18 < waldi> otavio: hopefully in the next two days
22:19 < bubulle> dunno if 2.6.31 is planned soon but I can bet it'll be out before the testing migration is done, and so on...
22:19 < maks> nno
22:19 < maks> please act more like kmuto
22:19 < maks> and integrate kernel preemptively
22:19 < waldi> bubulle: we are facing a "library" transition for each new version and have no usable way to handle that
22:20 < maks> we provide prebuild packages for the rc 
22:20 < otavio> maks: it is easier when you don't need to handle all arches
22:20 < otavio> maks: besides, coordination between team is a very nice thing to have
22:21 < otavio> maks: so intead of replying with a "no" would be nice to try to understand more deeply why bubulle has asked for it 
22:21 < maks> we need newer kernels in testing
22:21 < otavio> maks: btw it makes much more sense to wait until 2.6.x.y where y >= 2 to get it to sid 
22:21 < waldi> otavio: d-i managed to block transition several times during the lenny release cycle
22:22 < bubulle> I think the "why" is easy: we (and probably not only the D-I team ashes) need a kernel in testing that's newer than the kernel in lenny...:-)
22:22 < dilinger> yes, we do
22:22 < maks> not more in that road
22:22 < maks> s/more/more blocks/
22:22 < otavio> maks: it allows for faster migration to testing
22:22 < elmig> bubulle: rc1 will not be out soon, still on rc1 - will take months to get out
22:22 < elmig> (2.6.31)
22:22 < maks> otavio: your .2 question makes no sense, as any stable update can break abi
22:23 < maks> having older kernel in sid does *not* help testing migration
22:23 < maks> elmig: 2.6.31-rc2 is out
22:23 < otavio> maks: with that in mind, we can end asking for RM team to block it if we don't get support from kernel-team. This is very bad from "teamwise" POV  
22:24 < bubulle> <grin> apprently the current method doesn't either, sorry for being sarcastic
22:24 < otavio> maks: sure but most regressions would have been fixed in .2 or so and then the modules and like could have more chance to get into testing at that time
22:25 < maks> no
22:25 < maks> we caught the alpha ftbfs as we do multiple time due to upload
22:25 < maks> it is only fixed due to active work
22:25 < otavio> and why experimental couldn't do the same?
22:26 < otavio> afaik there're buildds for most arches now
22:26 < pusling> experimental autobuilders are crap
22:26 < waldi> otavio: experimental is for things which are _not_ supposed to be released
22:26 < bubulle> amen
22:27 < otavio> waldi: from my POV if kernel takes two mounths to be ready to testing it is *not* supposed to be released 
22:27 < otavio> waldi: so experimental fits quite well
22:27 < maks> why could d-i not use buildserver
22:27 < waldi> otavio: okay, we drop the stable ABI offer. this fixes this
22:27 < otavio> maks: because we all have real life and don't have time to do all that right now  
22:27 < otavio> maks: patches are more then welcome
22:28 < waldi> no more external modules
22:28 < waldi> and therfor no large transition
22:28 < otavio> waldi: makes no sense et all
22:28 < maks> external modules suck and should land in staging anyway
22:28 < maks> no makes sense
22:28 < waldi> otavio: sure, it does
22:28 < otavio> waldi: instead wait until it gets stabilized in exp and then we get the good from both world
22:29 < otavio> waldi: people wanting new kernel can take it from it
22:29 < waldi> this will not help
22:29 < bubulle> well, ok, I think we went all around that topic and apparently no real progress is happening
22:29 < waldi> you can't migrate things from experimental to unstable or testing
22:30 < otavio> maks: i'm not in place to argue about modules but I think we need to solve it all in some way 
22:30 < otavio> maks: from now, exp looks the best alternative
22:30 < bubulle> so, I propose we stop dscussing this right now, unless something constructive is proposed
22:30 < otavio> maks: please suggest a better one if you see it
22:30 < bubulle> the message then becomes: we should get the release team stance on this.
22:31 < bubulle> my personal POV being, from a quite far away point that we anyway need a new kernel version in testing and *not only* for D-I purposes....also because that would expose the new kernel to much more people
22:33 < maks> buildserver exposes to quite some people already as we get to know from bug reports
22:33 < bubulle> waldi, maks: by reading http://packages.qa.debian.org/l/linux-2.6.html I get the feeling that having 2.6.30-1 enter testing is not that complicated. Where am I wrong?
22:33 < maks> you need force hints bubulle 
22:33 < bubulle> maks: more specifically?
22:33 < maks> phil tried 2.6.29-2 the soft way and went nowhere
22:33 < bubulle> (for my own education and for future readers of this log)
22:33 < waldi> bubulle: this does _not_ show reverse deps
22:35 < bubulle> hmm, ok. that needs force hints....I have no idea why, probably because I'm dumb, but I guess I'm the only one in the world to not uderstand so let's no insist..:)
22:35 < waldi> bubulle: there are at least 3 _large_ packages depending on linux-2.6, they currently build around 2k binary packages with modules
22:36 < waldi> they need to be up-to-date and with the amount of changes in the kernel interface, about a half fails with every new version
22:36 < bubulle> otavio: about parted, you take care of the remaining build, you upload and then we're done, right?
22:36 < waldi> bubulle: as I said, this is a _large_ "library" transition
22:36 < maks> bubulle: this gives already more realistic picture http://release.debian.org/migration/testing.pl?package=linux-2.6
22:38 < bubulle> maks: thanks. That helps understanding. You know, all these things are maybe easy to guess for people who constantly deal with large transition, but they aren't for me...and I have the weakness to think that I'm not alone..:)
22:39 < bubulle> otavio: wrt release preparation, this is all what we have to discuss about or are there other things?
22:39 < otavio> bubulle: for now; yes
22:39 < otavio> bubulle: all the rest depends on that
22:39 < bubulle> ok....before we move the kfreebsd stuff you wanted to discuss with slackydeb, what about console-setup things?
22:40 < bubulle> I think we're basically not ready to release with it (see fjp objections)
22:40 < otavio> i agree
22:41 < otavio> alpha1 will need to stay with kbd-chooser
22:41 < otavio> but we could target it for very soon now
22:41 < bubulle> in the meantime, I'll try to work with anton and youpi to have a more acceptable c-s
22:41 < otavio> it looks like last upload has fixed many of those problems and that you and Samuel are looking at the missing layouts and like 
22:41 < bubulle> I still think that we should reduce the complexity of the keymap questions
22:42 < otavio> bubulle: yes
22:42 < bubulle> I just prepared a netboot image with latest console-setup. I'll try to announce it in the mailing list, in the hope that it will be throughly tested
22:43 < bubulle> Samuel and I are not hunting the missing layouts: I try to push my idea of making a selection of layouts to propose instead of offering all possibilities
22:43 < bubulle> still, this is work in progress
22:43 < otavio> bubulle: i agree with you that we ought to offer only supported languages related keyboards
22:44 < otavio> bubulle: i saw the thread
22:44 < bubulle> Not much progress, on the other hand, about dhcp patch for netcfg
22:44 < otavio> bubulle: but i also think it could be done using the Samuel's patch for it
22:44 < otavio> bubulle: his script
22:44 < bubulle> (back on c-s) yes...and I bet that fjp would agree too (which would have great importance to me)
22:44 < otavio> bubulle: using a whitelist for the languages to add 
22:45 < otavio> bubulle: but using his script to get the variants
22:45 < otavio> bubulle: what do you think?
22:45 < bubulle> otavio: yes...and no. We probably need to manage some specificities manualy. Example: what Brazilian keymap should we offer?
22:46 < otavio> bubulle: intl and native
22:46 < otavio> bubulle: abnt2 and us_intl
22:46 < bubulle> we can't offer all variants...too many of them for many layouts.....but we can't also only offer exactly one variant for each layout as, for some of them (or for some languages) we need more than one layout/variant
22:47 < bubulle> otavio: yes....that's my point: we need input by Those Who Know, ie the people from the relevant countries/languages communities
22:47 < bubulle> so, manual selection, just as we did with kbd-choose
22:47 < bubulle> chooser
22:47 < otavio> bubulle: humm; ok
22:47 < otavio> bubulle: from that pov i agree
22:47 < otavio> bubulle: yes
22:47 < bubulle> this is not that a big deal to handle: after all that's kind of the job of the i18n person..-)
22:48 < otavio> bubulle: so i'd say to you to push your selection and then we see if people screeam
22:48 < otavio> bubulle: hah ok
22:48 < bubulle> yup. Hopefully, I'll have a working patch (if only Samba folks stopped releasing new versions...:-))
22:49 < bubulle> About DHCP patch for netcfg, I haven't seen much feedback, indeed. And, as luk is not around, I think we can currently definitely rule it out for alpha1
22:49 < otavio> ok
22:49 < otavio> well, not really
22:50 < otavio> looking forward about kernel issue it looks like we need still few weeks for it 
22:50 < otavio> and then netcfg can be sorted out
22:50 < otavio> i can look at his patch and check if i can fix the remaining issue
22:50 < otavio> issues
22:51 < bubulle> yup...we'll see
22:51 < bubulle> I think that it's OK now if you want to discuss kfreebsd stuff with slackydeb 
22:51 < slackydeb> bubulle: here
22:51 < bubulle> I haven't other improtant things to discuss
22:52 < bubulle> and it would be good to have this topic inside the "official" meeting
22:52 < bubulle> up to you guys
22:52 < slackydeb> about dhcp, please don't enable busybox one without letting the opportunity to use the old one
22:52 < slackydeb> please
22:52 < otavio> slackydeb: exactly what i we going to ask
22:52 < slackydeb> I didn't port it yet
22:52 < otavio> slackydeb: busybox doesn't work?
22:52 < slackydeb> what part :)
22:52 < slackydeb> ?
22:53 < slackydeb> otavio: at the moment there is a known bug in the ash job control
22:53 < slackydeb> otavio: because of this at the moment I disabled debug shell
22:54 < slackydeb> otavio: (yes, it was a debug nightmare)
22:54 < otavio> slackydeb: isn't it solved in 1.14?
22:54 < slackydeb> otavio: I didn't try it
22:54 < slackydeb> otavio: waldi this mornig told me that a new upload is pending
22:54 < slackydeb> otavio: but where is the debian busybox vcs???
22:54 < otavio> slackydeb: his personal dir inside of d-i svn
22:55 < otavio> waldi: about busybox i saw you've enabled new applets in udeb
22:55 < otavio> waldi: and them doesn't look to being used sha* for example 
22:55 < waldi> otavio: where?
22:55 < waldi> i thought i only did that for the others
22:55 < otavio> waldi: at least from the commit it looked to be enabled
22:55 < otavio> waldi: just take a look
22:56 < otavio> waldi: i might have got confused by the diff but anyway
22:56 < waldi> +# CONFIG_SHA256SUM is not set
22:56 < otavio> waldi: so far so good
22:56 < otavio> waldi: so no problem
22:56 < slackydeb> otavio: thanks, I found that vcs :)
22:57 < otavio> slackydeb: np
22:57 < otavio> slackydeb: besides that, how is kfreebsd going? what current state?
22:57 < slackydeb> otavio: I need some help to (propose to) merge more things to trunk
22:57 < otavio> slackydeb: yes; let's try to keep merging stuff was they get ready
22:58 < slackydeb> otavio: I didn't manage to split the rootskel-bootfloppy binary building from rootskel
22:58 < otavio> slackydeb: so your branch do not diverge a lot
22:58 < otavio> slackydeb: you didn't? it should be easy to do
22:58 < slackydeb> otavio: I have a lot of work in rootskel, it is stabilizing, but I can't merge because of this
22:58 < otavio> slackydeb: got it
22:58 < otavio> slackydeb: but no other stuff ready?
22:58 < otavio> slackydeb: installer?
22:59 < slackydeb> otavio: yes, it is... but dh_something complained...
22:59 < slackydeb> otavio: in the installer directory?
22:59 < otavio> slackydeb: yes; please check if there's anything ready and propose
22:59 < slackydeb> otavio: installer/build/config is stabilizing, I could propose it
22:59 < otavio> slackydeb: you can catch me at irc and we check it together with pastebin
22:59 < bubulle> if I'm correct, slackydeb you'll be at Debconf, right?
22:59 < otavio> slackydeb: before you propose to ml
23:00 < slackydeb> bubulle: no
23:00 < bubulle> arg
23:00 < bubulle> I thought so seeing the commit of the gsoc talk
23:00 < slackydeb> bubulle: :)
23:00 < bubulle> but I guess the slides will be presented by Arthur in a general talk about gsoc projects, then
23:00 < otavio> well
23:01 < slackydeb> otavio: and there is another little help needed...
23:01 < otavio> from my pov i think we can close the meeting
23:01 < bubulle> sad, because that would have been a great opportunity to work with otavio (not to mention me: I'm dumb) on merge
23:01 < otavio> slackydeb: just say it 
23:01 < slackydeb> otavio: but I din't find time to try this
23:01 < slackydeb> otavio: busybox config split
23:01 < bubulle> I agree we can close the meeting.....ad I forgot opening it with Meetbot that will be easy
23:01 < bubulle> ;p
23:01 < bubulle> End of meeting...:-)