People who speak below:

bubulle   : Christian Perrier
joeyh     : Joey Hess
fjp       : Frans Pop
aba       : Andreas Barth
vorlon    : Steve Langasek
ths       : Thiemo Seufer
Kamion    : Colin Watson
zinosat   : Davide Viti
attilio   : Dr. Attilio Fiandrotti
sc        : Stefano Canepa
stappers  : Geert Stappers

Hours are European Time (UTC +0200):


21:49 < bubulle> OK, first round to know who's already around
21:49 < joeyh> hi
21:49  * fjp is
21:50 < joeyh> is this the pre-meeting meeting? :-)
21:50 < [-oskar-]> :)
21:50 < aba> hi
21:50 < bubulle> joeyh: nope...just a way to avoid the "I'm here" rush at the beginning
21:51  * vorlon pretends to be here
21:51 < fjp> Ah, was that for the meeting. Hmm, not sure I can make it...
21:51  * aba will disappear in time for the meeting
21:52 < zinosat> hi
21:52  * bubulle has assigned a time schedule for the 3 propsoed topics
21:52 < bubulle> see wiki
21:53 < ths> I'm also here.
21:54 < Kamion> I'm here, not sure for how long
21:56 < bubulle> Kamion: should be less than 50mins I hope
21:56 < zinosat> fjp, got in touch with Attilio?
21:59 < bubulle> OK, anyone else around?
22:00 -!- bubulle changed the topic of #debian-boot to: D-I devel team meeting in progress | Topic: Status of beta release
22:00 < bubulle> Here we are....
22:00 < bubulle> http://wiki.debian.org/DebianInstallerEtchBeta1Prep
22:00 < joeyh> well, I wish I could announce some images to test for the beta, but that's not happened
22:00 < joeyh> things are reasonably dire as of yesterday when we discovered many new problems
22:01 < bubulle> listed in the wiki?
22:01 < joeyh> yes, there and at DIToday
22:01 < bubulle> http://wiki.debian.org/DebianInstallerToday
22:01 < joeyh> fixes for most are in progress, but have to get to testing so it will take a while
22:02 < bubulle> secure-apt on its way as I've seen
22:02 < bubulle> adduser bug should be soon fixed
22:02 < bubulle> maybe request for medium urgency uploads?
22:03 < attilio> hi, i've finally arrived home
22:03 < joeyh> I'm hinting most of these in at high urgency, it still takes a while
22:04 < bubulle> secure-apt, adduser, deboostrap are being worked on by their maintainers, right?
22:04 < joeyh> debootstrp is fixed as of dinstall
22:04 < joeyh> adduser doesn't affect the beta
22:04 < bubulle> because not in testing?
22:04 < Kamion> bubulle: right
22:04 < joeyh> secure-apt I've asked mvo to try to fix sooner than he'd planned
22:05 < joeyh> preseed will be fixed after britney if I can get ahold of jvw
22:05 < bubulle> what about the pcmcia pb?
22:05 < joeyh> pcmcia-cs has to autobuild and I've hinted it to high (hope that's ok, kamion)
22:05 < joeyh> (also hope it's well tested..)
22:06 < fjp> One thing that has happened is the installer/manual split. I'm uploading new packages for the manual tonight. Should make things a bit easier in the future.
22:06 < Kamion> it's tested
22:06 < Kamion> note careful wording :-)
22:06 < joeyh> in d-i too?
22:06 < vorlon> debootstrap still needs an upload for the libsigc change?
22:06 < Kamion> joeyh: yes
22:06 < bubulle> joeyh: so if everything goes well, maybe some images to test in a few days?
22:06 < joeyh> no, debootstrap had 2 nmus and as of the last I think it will work
22:06 < vorlon> oh, ok
22:06 < joeyh> it will need to be synced to testing tho
22:06 < joeyh> maybe around the weekend
22:07 < bubulle> which arches?
22:07 < joeyh> parted is also ready to sync into testing
22:07 < Kamion> the pcmcia-cs thing has been around for months apparently not-much-noticed - perhaps we need to improve our testing coverage there
22:07 < joeyh> looks like all arches now, assuming alpha's autobuild failure doesn't repeat
22:07 < bubulle> wow
22:07 < fjp> Kamion: For that you need the hardware.
22:07 < joeyh> ia64 got worked around
22:07 < bubulle> do we have testers for all arches?
22:07 < Kamion> fjp: indeed, I don't necessarily mean autotest
22:08 < sc> Kamion: I should test on a laptop
22:08 < joeyh> bubulle: not really
22:08 < joeyh> yes, we also had the usb keychain installs broken for a while w/o nyone noticing
22:08 < joeyh> (while = 2 months)
22:08 < bubulle> we probably need to put more attention on all styles of "exotic" installs
22:09 < joeyh> it's a beta, it doesn't have to be perfect but it would be nice to avoid most of the worst issues
22:09 < bubulle> agreed
22:10 < bubulle> it's likely to receive quite wide testing because it's the first etch installer
22:10 < joeyh> anyway, I wanted to talk about what to do about ongoing development while the beta is mturing
22:10 < joeyh> are people feeling limited by the impending beta?
22:10 < joeyh> I think I am ;-)
22:10 < bubulle> fine. anything else for the beta release topic so that we can close it?
22:11 < joeyh> would like to upload the new apt-setup but unsure about breaking unstable installs just now
22:11 < fjp> That's mostly the time you're spending on tracking things though, not real blockers?
22:11 < Kamion> joeyh: exactly what I was about to say
22:11 < Kamion> not been feeling limited apart from that
22:11 < bubulle> Well, let's close the beta release topic then....and open another one
22:11 < joeyh> having unstable installs working to test things before they hit testing is important right now
22:12 < joeyh> so I don't know if there's much we can do aside from getting testing in a working state
22:12 < bubulle> D-I devel team meeting in progress | Topic: ongoing development
22:12 < bubulle> oops
22:12 < joeyh> bubulle: it's basically the beta topic really..
22:12 -!- bubulle changed the topic of #debian-boot to: D-I devel team meeting in progress | Topic: ongoing development
22:12 -!- bubulle changed the topic of #debian-boot to: D-I devel team meeting in progress | Topic: beta release and ongoing development
22:12 < bubulle> right?
22:13 < joeyh> yeah, although we may have exhausted the topic now ;-)
22:13 < Kamion> the other thing I'd like to do is basically summarised as "switch to udev and see what breaks"
22:13 < bubulle> well, let's eat 5 mins to the next topic
22:13 < joeyh> f you feel limited talk to me or fjp and we'll see what we can come up with I guess
22:13 < bubulle> joeyh: you have up to 20:20
22:13 < Kamion> but I'm almost certain stuff will break there, and not a lot to be done except find out what it is and squash it
22:14 < joeyh> someone could implement the d-i feature I dearly want of pulling udebs from a supplimental repo and then we could test things more easily
22:14 < Kamion> we have to do it though, we're rapidly running out of devfs time
22:14 < fjp> I would say hold off to the beta, switch daily builds to etch_d-i for a while and seriously break sid_d-i?
22:14 < bubulle> seems fair to me 
22:15 < bubulle> this will hurry us for the beta
22:15  * joeyh expects it will just work, kinda like his apt-setup code
22:15 < Kamion> in my recent testing with udev, with the exception of #334584 (which didn't seem to have particularly serious consequences for me in practice), things seemed to be basically working; expected breakage is more in areas like exotic partitioning
22:15 < bubulle> and give at least 2 main goals for post-beta1: apt-setup and udev
22:15 < fjp> We could even build kernel udebs from exprimental kernels?
22:16 < Kamion> come on, there must be post-beta1 goals I haven't been working on ;)
22:16 < bubulle> graphical installer?
22:16 < Kamion> oh yeah, duh
22:16 < fjp> What about that umount -a problem with initramfs?
22:16 < Kamion> fjp: we don't have to switch to initramfs for udev, although we could ...
22:16 < joeyh> it was still there as of oldenburg
22:17 < fjp> No, but we do need it for the next kernel releases.
22:17 < ths> Kamion: Some architectures can't yet.
22:17 < Kamion> I should have a look, I was thinking of switching Ubuntu to initramfs anyway to experiment with memory savings
22:17 < Kamion> fjp: not in the first stage we don't
22:17 < Kamion> fjp: the umount -a / initramfs problem is only in the first stage with busybox, and initrd's still fine there even with new kernels
22:17 < Kamion> ths: can't ...?
22:18 < fjp> Kamion: Right.
22:18 < ths> Kamion: 2.4 kernels, limited bootloader support.
22:18 < bubulle> hmmmm, discussion slowing down....anything else to ad don this topic? I won't extend it again?
22:18 < Kamion> ths: oh, right, sure, how the d-i rootfs images are built is arch/kernel-specific anyway
22:19 < Kamion> but yes, we're derailing a bit
22:19 < fjp> ths: We keep non-udev support for older kernels.
22:19 < fjp> bubulle: Next?
22:19 < bubulle> yep
22:19 -!- bubulle changed the topic of #debian-boot to: D-I devel team meeting in progress | Topic: graphical installer status
22:20 < bubulle> fjp: up to you and zinosat 
22:20 < fjp> Loads of great progress there :_)
22:20 < fjp> :-)
22:20 < zinosat> yes
22:20 < fjp> The last pieces we need to fully build for the archive are in new or building.
22:20 < attilio> and more still to come ;)
22:20 < Kamion> fjp: that's truly awesome
22:21 < bubulle> fjp: can the gtk-miniiso be integrated in regular builds just after beta?
22:21 < zinosat> we need cdebconf-gtk built with Alistair's libs
22:21 < zinosat> that's really the last important thing
22:21 < fjp> zinosat: Alastair's libs are in NEW. So waiting for that.
22:21 < bubulle> fonts?
22:21 < fjp> Will try to build from them as soon as they're in the archive.
22:22 < zinosat> fjp, yep. after that only some fonts will have to be udebbed
22:22 < fjp> For fonts I think we first need some expert advise...
22:22 < bubulle> any contacts with font packages maintainers?
22:22 < fjp> Not yet.
22:22 < joeyh> what kind of memory footprint are we looking at for the grphical installer?
22:22 < zinosat> fjp, yes, we should look for somebody with the knowledge and the time
22:22  * bubulle can try pushing Jaldhar
22:23 < attilio> who's going to contact fonts packagers for help ?
22:23 < Kamion> it's probably time to bring debian-gtk-gnome@lists into the discussion if you haven't already ...
22:23 < bubulle> dunno if he has some deep font knowledge bu he maintains several of them
22:23 < fjp> I was thinking of doing an announcement to d-d-a after the last bits are in place and asking for help in that.
22:23 < bubulle> OK, recorded
22:23 < zinosat> joeyh, I was looking for another minimal installer to do size  comparison
22:23 < Kamion> -gtk-gnome is a list where plenty of the relevant developers hang out
22:23 < fjp> joeyh: initrd size is about 25MB
22:24 < bubulle> usability issues should be looked at, at some moment
22:24 < joeyh> so maybe 64 mb?
22:24 < fjp> I'll have a look at memory usage during install soon.
22:24 < zinosat> Kamion, how should be modified as to build agains directfb-gtk?
22:24 < fjp> That's optimistic, but maybe.
22:24 < zinosat> Kamion, cdebconf..
22:25 < fjp> attilio: What are your plans for the frontend?
22:25 < bubulle> what about 2nd stage, btw?
22:25 < attilio> next days i plan to work a lot on it
22:25 < zinosat> Kamion, changing MODCFLAGS=`pkg-config --cflags gtk+-2.0` in the Makefile?
22:25 < attilio> i'm working on some aspect and usability issues
22:25 < Kamion> zinosat: if Alistair added a suitable pkg-config file, it's a trivial and fairly obvious modification to MODCFLAGS and MODLDFLAGS
22:25 < joeyh> bubulle: one of the reasons I want to get rid of second stage ..
22:25 < Kamion> zinosat: if not, somebody should get him to add one
22:25 < zinosat> Kamion, right
22:26 < attilio> tomorrow i should be able to post on d-boot a complete list of improvements i'm planning to do.
22:26 < fjp> attilio: I'll send a mail to the list with some usability things I've noticed.
22:26 < fjp> Cool
22:26 < attilio> fjp, collecting usability issues is really useful.
22:26 < bubulle> fjp: about usability we have to take care that templates are actually written with some interface-isms
22:26 < Kamion> second stage> yeah, I doubt it's worth trying to do all the same changes to debconf and getting a second-stage framebuffer up and such ...
22:26 < fjp> Kamion: Yes he did. Everything should be in /usr/lib/gtk-directfb or something like that.
22:27 < bubulle> Kamion: which means that I really shoudl take care of first stage password setting..:-)
22:27 < fjp> Kamion, joeyh: Any chance of getting cdebconf to support the notion of columns?
22:27 < Kamion> bubulle: let's not rush into it - I think a lot of the same questions arise for passwd and for the rest of base-config
22:27 < Kamion> fjp: in what context?
22:27 < joeyh> it would probably need a new templte type
22:28 < joeyh> or some extensions, it's certianly doable..
22:28 < Kamion> wasn't there a CONTAINER type in debconf once?
22:28 < fjp> Kamion: language selection has 2 columns, partman's main screen also
22:28 < joeyh> Kamion: in the spec yes
22:28 < Kamion> fjp: plugins, damnit :-)
22:29 < Kamion> the only difficulty is rearranging cdebconf's build system so that it's possible to build out-of-tree plugins
22:29 < attilio> on the two columns language chooser: it should be realizable without modifying the templates.
22:29 < Kamion> other than that it should be possible for localechooser to ship a suitable set of gtk widgets itself if it needs unique widgets
22:29 < attilio> on the partman's main menu: here maybe an hack could be inserted inside the frontend to use a tree instead of a list.
22:29 < Kamion> I think the localechooser and partman widgets probably want to be somewhat different
22:29 < fjp> attilio: Yes, but only by creating question-specific handler, which is not very nice in the long run.
22:29 < Kamion> attilio: have you looked at the plugin stuff at all? I really think it's better than hacking the generic frontend
22:30 < bubulle> so...we're running into details now...so post-meeting....
22:30 < zinosat> oh... ports: powerpc is being worked on (by eddyp and fjp). anyone interested for other arches? once cdebconf-gtk will be available for all arches it's gonna be just a matter of tweaking the build system
22:30 < Kamion> attilio: remember that eventually we want to use cdebconf outside the installer too :-)
22:30 < attilio> kamion, not yet, ATM i'm hacking the frontend but providing proper comments to explain why a hack is there and what does it does
22:31 < bubulle> zinosat: other than i386 and powerpc, what arches could be OK with the gtk installer? amd64, I guess
22:31 < attilio> and i always keep and eye to the possibility of using the GTK frontend outside the d-i and with multiple question
22:31 < zinosat> bubulle, don't know honestly...
22:31 < fjp> amd64 is not a problem. sparc comes to mind, but may have a difficult framebuffer
22:32 < joeyh> well you could run it on any arch if you had an open DISPLAY on yout laptop
22:32 < joeyh> and a boot parameter
22:32 < joeyh> well, probbly not really ;-)
22:32 < bubulle> joeyh: well, realistically...:-)....
22:32 < fjp> with gtk-directfb and without networking set up?
22:33 < fjp> Anyway, I'm very happy with the progress we've made and thanks to attilio and zinosat for their work.
22:33 < bubulle> OK. we're nearly over the topic now and we have a pretty good picture, I think
22:33 < zinosat> fjp, thak yoursel too :)
22:34 -!- bubulle changed the topic of #debian-boot to: D-I devel team meeting in progress | Topic: Organising a d-i BSP
22:34 < attilio> i agree: having an automated system for building isos is agreat step ahead
22:34 < bubulle> let's move...
22:34 < joeyh> ok, this is just an idea I had.
22:34 < joeyh> I know that we have a lot of old bugs, some even with patches that we put off applying
22:35 < joeyh> so I spent some time this week organising them and have gotten some views of all the d-i bugs tht are fairly useful
22:35 < joeyh> http://wiki.debian.org/DebianInstallerBugs
22:35 < joeyh> idea would be to try to go over as many of them as we can
22:35 < bubulle> in an organized manner (with a schedule and so on)?
22:36 < bubulle> BSP week-end?
22:36 < fjp> Main question that comes to mind is: who's the "we"?
22:36 < joeyh> yeah, that's the questiin
22:36 < bubulle> probably the people attending this meeting..:-)
22:36 < joeyh> a lot of these will probably come down to seeing it it still reprocuces
22:36 < joeyh> I think that could be distributed out a bit
22:37 < bubulle> so it needs someone to organize the stuff
22:37 < fjp> What we really need is to get ppl like Thomas Hood, joshk, Anton, ... involved again.
22:38 < bubulle> hmmm, noone volunteering to organize the first D-I BSP?
22:38 < joeyh> another related thing is installation reports
22:38 < joeyh> I thought that geert or someone was planning to close a lot of the old ones
22:39  * stappers stays quiet
22:39 < fjp> Yes, it was something for Oldenburg, but we did not really get around to that.
22:39 < fjp> And with time spent on GTK, I did not really feel like taking it on.
22:40 < joeyh> one of the views I worked up sorts bugs by architecture, so we could try to get some more porters involved too
22:40 < bubulle> last meeting we agreed to close all install reports before beta3
22:40 < joeyh> all those are usertagged as old-report now BTW, and excluded from these pages
22:41 < bubulle> so should be easy to close in a big bang
22:41 < joeyh> anyway yes I would like to organise it, but the beta has to be my first priorit
22:41 < bubulle> "BSP two weeks after the beta"
22:41 < joeyh> it's probably something for just after the beta release
22:41 < bubulle> s/two/one
22:42 < bubulle> OK, well, we seem to run out of energies here....:-)
22:42 < fjp> Do we announce it on d-d-a?
22:43 < joeyh> yeah, most likely
22:43 < joeyh> ok, one more thing about this..
22:43 < bubulle> agreed as well
22:43 < joeyh> there are several packagtes like mklibs and busybox that show up on the page as they're maintained by debian-boot
22:43 < joeyh> but I've noticed not all their bugs actually affect d-i
22:43 < joeyh> cdebconf is another one
22:43 < joeyh> not sure what to do about that
22:44 < fjp> Well, as we are the maintainers, we should probably care about them...
22:45 < fjp> busybox is mainly waldi; cdebconf mainly joeyh and Kamion; mklibs waldi and now a bit me I guess :-P
22:45 < vorlon> joeyh: do you want to give people here the url for that per-arch bug view, so it might wind up in any minutes from the meeting?
22:46 < joeyh> it's a long url, linked to from http://wiki.debian.org/DebianInstallerBugs
22:46 < fjp> http://bugs.debian.org/cgi-bin/pkgreport.cgi?maint=debian-boot@lists.debian.org;ordering=archfirst
22:46 < joeyh> it's also possible to do views that only list bugs that affect one arch
22:46 < bubulle> well, I think we can barely close the meeting
22:46 < fjp> We should probably set that up as default selections too.
22:47 < bubulle> with the trandtional meeting question: when to do the next one?
22:47 < joeyh> the other one (by category) is currently the default view
22:47 < joeyh> 2 weeks on a weekend isn't it?
22:47 < zinosat> bubulle, some weekend (sat) next month
22:47 < fjp> BSP then probably...
22:47 < bubulle> Sat Now 19th
22:48 < bubulle> for the Nov meeting
22:48 < fjp> Sounds OK
22:48 < bubulle> hour, I don't remember what I proposed
22:48 < zinosat> bubulle, iirc weekends were earlier
22:49 < bubulle> 16:00UTC is what I finally proposed
22:49 < bubulle> OK, let's close that meeting. Thanks to everyone involved
22:50 < bubulle> back to work^W bed now....