From Fedora Project Wiki
< Extras | SteeringCommittee
Summary
Present from FESCo: thl, warren, jeremy, scop, skvidal, Anvil, ensc
Important topics:
- Mass rebuild of Extras for FC5
- 69 packages still need a rebuild; ensc has a script that will file bugs for those
- we'll remove all the old packages that are obsoleted by a newer version when branching FE5. Additionally, in the future we'll only keep the newest version of a package in the devel tree and remove the older ones directly -- just as it is done in rawhide. For supported distros the nothing changes (e.g. the newest package and one older package)
- Try to fix broken deps in FE5 before FC5 is published; If that's not possible we'll consider removing packages with broken deps
- Split fedora-extras-list
- warren> I'm moving ahead with the transition of the fedora-extras list thing within the next week
- Kernel module standardization
- some test builds for FE5 worked wine; we'll see what happens if people start to use those
- building for ppc64 not that easy; thl will look at it
- EOL Policy for FE
- thl plans to bring that forward after FC5
- Extras to handle multilib
- Core is working on a proper solution that we might adopt for Extras in the longer term. Until then we'll ignore multilib mostly. Special exception: wine. wine.x86_64 doesn't make that much sense atm and many people requested wine.i386 in the x86-64-repo. We'll allow that by manually copying it (and the needed deps) over. Request for copying via FC[3-5] Status-pages in the wiki. No other exceptions should slip in, otherwise it will get unmanageable with this manual way.
Full Log
18:59 --> | Sopwith (Undisclosed) has joined #fedora-extras
19:00 --- | thl has changed the topic to: FESCo Meeting in progress
19:00 < thl> | Hello everyone
19:00 < skvidal> | hi
19:00 < thl> | who's around?
19:00 < skvidal> | I am
19:00 < jeremy> | hi thorsten! :)
19:00 < warren> | hi
19:01 --- | thl has changed the topic to: FESCo Meeting in progress -- Mass rebuild of Extras for FC5
19:01 < roozbeh> | warren: i have went through the whole thing, i know. but lemme test and see what happens. i will contact you again if i found you were wrong ;)
19:01 < thl> | okay, current status
19:01 < thl> | 69 packages still need a rebuild
19:01 < thl> | I just posted about some orphans to extras-list
19:01 --> | JSchmitt (Jochen Schmitt) has joined #fedora-extras
19:01 < thl> | is it okay for everybody if we remove them before FE5 is live?
19:01 < ensc> | thl: my mass-bz script seems to work...
19:02 < ensc> | (finished 1 minute ago ;) )
19:02 < thl> | hey, ensc, that'sgreat
19:02 < warren> | thl, i'm rebuilding my packages
19:02 < thl> | warren, thx
19:02 < roozbeh> | is owen taylor's package still not rebuilt? it is practically orphaned, since he told me he doesn't have the extras build env anymore
19:03 < roozbeh> | libuniname or sth
19:03 < thl> | roozbeh, seems someone build it afaics
19:03 < roozbeh> | ok, nice to hear
19:03 < thl> | I take the silence as "yes, remove those if no one complains on the list in the next 24 hours"
19:04 < jeremy> | thl: sounds good to me
19:04 < roozbeh> | thl: lovely
19:04 < thl> | k
19:04 < thl> | for the 69 still not rebuild
19:04 <-- | roozbeh has quit ("Leaving")
19:04 < thl> | ensc, can you test you script with those?
19:04 < thl> | some bugs are files already
19:05 < thl> | look at http://www.fedoraproject.org/wiki/Extras/FC5Status before you run it
19:05 < ensc> | thl: do you have a wording for the bugzilla comment?
19:05 < thl> | no
19:05 < thl> | just write down some line that might fit
19:06 < thl> | and link to the initial rebuild announcment
19:06 < thl> | next one regarding that topic
19:06 < thl> | jeremy said branching of extras should happen soon
19:06 < ensc> | do we have a tracker bug?
19:07 < thl> | ensc, yeah, FE5TRACKER iirc
19:07 < jeremy> | thl: yeah, I'm looking into what all is needed to have the branching not be a disaster now
19:07 --> | scop (Ville Skytta) has joined #fedora-extras
19:07 < thl> | there was one more thing regarding FE5 that came up:
19:08 < thl> | shall we remove all the old packages that are opsoleted by a newer version?
19:08 < Anvil> | scop : hello
19:08 < thl> | I think we should
19:08 * | scop blushes, late again
19:08 < warren> | thl, yes
19:09 < thl> | anyone around who thinks that's a bad idea ?
19:09 < jeremy> | sounds good to me
19:09 < Sopwith> | Yes, we should.
19:09 < thl> | jeremy, can you do that yourself with repoquery?
19:10 < thl> | btw, I'm wondering if we should use a similar tactic for the extras/devel tree in general
19:10 < jeremy> | thl: I can figure something out
19:10 < thl> | e.g. just like rawhide
19:10 < thl> | only the newes package is available, old ones get deleted immediately
19:10 < thl> | less space on the servers and easier to manage
19:10 < thl> | jeremy, thx
19:11 < thl> | other opions on that tactic for extras/devel/ ?
19:11 < jeremy> | I'm not against it
19:11 < thl> | s/opions/opinions /
19:12 < warren> | Does that make repocreate and repoview faster?
19:12 < scop> | works4me for devel
19:12 < thl> | warren, maybe, not sure
19:12 < scop> | probably somewhat
19:13 < thl> | okay, so we change that in devel after the FE5 tree was created?
19:13 < thl> | skvidal, can you change the script accordingly?
19:13 < thl> | Or is that a complicated?
19:14 < scop> | not complicated, I can take care of it
19:14 < thl> | scop, thx
19:14 < thl> | okay
19:15 < thl> | are there any other things regarding Mass rebuild/FE5 that need discussing?
19:15 < thl> | otherwise it might be to late
19:15 < warren> | Are we removing things not rebuilt, or filing bugs?
19:15 < thl> | warren, filing bugs
19:16 < thl> | k, moving on
19:16 < warren> | I'm moving ahead with the transition of the fedora-extras list thing within the next week
19:16 < thl> | :)
19:16 < thl> | thx warren, that was on my "ask warren list"
19:16 --- | thl has changed the topic to: FESCo meeting -- Kernel module standardization
19:17 < thl> | scop, any news?
19:17 < thl> | building the modules worked fine afaics
19:17 < scop> | the test builds look pretty much ok
19:17 < thl> | k
19:17 < thl> | I'll rebuild some other modules with the new same in the next days
19:18 < thl> | we'll see what happens if people start to use those
19:18 --- | thl has changed the topic to: FESCo meeting -- EOL Policy for FE
19:18 < thl> | any news on this one?
19:18 < scop> | I added ppc64 and ppc64iseries to lirc-kmod just to test, but the build is stuck in buildsys
19:18 < scop> | (sorry, still on kmods)
19:18 < thl> | scop, np
19:18 < scop> | https://bugzilla.redhat.com/185387
19:19 < scop> | I'm done now, please proceed :)
19:19 < thl> | warren, can you poke dcbw to look at #185387
19:19 < thl> | kernel-modules for ppc64 sound like a good idea
19:19 < thl> | ppc64iseries seems no that important to me for now ;-)
19:20 < warren> | yeah
19:20 < thl> | that can wait some weeks if that's more complicated
19:20 < thl> | k
19:20 < thl> | so back to EOL
19:20 < jeremy> | ppc64 in general may be complicated :-/
19:20 < warren> | would you really be using lirc on ppc64*?
19:20 < scop> | no
19:20 --- | thl has changed the topic to: FESCo meeting -- kmods and ppc64
19:20 < scop> | O
19:20 < thl> | jeremy, why?
19:20 < scop> | oops
19:20 < warren> | then why build it there?
19:21 < scop> | I have repeatedly mentioned on the extras list that I'm testing the infrastructure with these builds
19:21 < scop> | and that I will remove the packages from the repo before FC5
19:21 < jeremy> | thl: for core, all ppc64 builds happen in a full ppc64 buildroot. we don't do a ppc64 release, and I'm not 100% sure that things will just work in a mixed env
19:22 * | thl thinks about that for a moment
19:22 < warren> | We don't publish FC5 ppc64
19:22 < thl> | we'll probably have to look at that closer and fix it after FE5
19:22 < warren> | there is no easy way of making a FC5 ppc64 buildroot
19:22 < thl> | not nice, but that's life
19:23 < warren> | I personally wouldn't worry about ppc64 kernel modules
19:23 < thl> | jeremy, any idea how that could be solved without a complete ppc64 tree?
19:23 < warren> | ppc itself is less than 1% of our users, and they will continue to shrink
19:23 < thl> | I have access to a ppc64 machine with plague on it
19:23 < thl> | I'll try to take a look
19:24 < jeremy> | thl: it might work. it's worth trying, but I just don't know if it will work
19:24 < thl> | maybe I can get it running somehow
19:24 < jeremy> | and if it doesn't, I'm not sure if there's much we can do :-/
19:24 < thl> | jeremy, k, that's life
19:24 < thl> | okay, anything else?
19:24 --- | thl has changed the topic to: FESCo meeting -- EOL Policy for FE
19:24 < thl> | okay, back to this one
19:25 < thl> | I plan to bring that forward after FE5
19:25 < thl> | there was to much todo in the last weeks
19:25 < thl> | Sorry for that
19:25 < warren> | Did we decide on anything regarding this topic earlier? I'm fuzzy.
19:25 < thl> | no, we didn't yet
19:25 < thl> | the only thing we said was:
19:26 < thl> | No new branches for FE3 in cvs
19:26 < thl> | but they are created nevertheless iirc
19:26 < thl> | no big deal
19:26 < warren> | I personally want one server utilty that I use added to FE3...
19:26 < warren> | but never had a chance to package it yet
19:27 < thl> | okay, then let's ignore that for now
19:27 --- | thl has changed the topic to: FESCo meeting -- Encourage Extras reviews
19:27 < thl> | any new ideas?
19:27 < warren> | heard any progress from Hans on the security thing?
19:27 < warren> | thl, nothing new, only need to improve documentation, no different than before. This is largely my responsibility.
19:28 < thl> | warren, no, didn't hear anything from hans
19:28 < thl> | I'll try to drive that forward together with the EOL thing
19:29 --- | thl has changed the topic to: FESCo meeting -- Broken deps report
19:29 < thl> | mschwendt not around
19:29 < thl> | let's skip it
19:29 < warren> | he posted stuff to the list
19:29 < warren> | like we agreed last week, broken deps are a bigger problem than not rebuilt packages
19:29 < thl> | yeah, but I think it needs a bit imporving
19:30 < thl> | and it should run on one of our severs automatically
19:30 * | warren goes to poke the owner of at-poke
19:30 < thl> | yeah, there are still some broken deps
19:32 < thl> | do we want to step up to fix those?
19:32 < thl> | file bugs today, if no one steps up until saturday to fix the deps
19:33 < thl> | then sponsors can fix it
19:33 < warren> | I'd say just go ahead and fix them
19:33 < warren> | file bugs if they aren't obvious
19:33 < thl> | anyone interested in doing the actuall work?
19:34 <-- | Foolish has quit (Read error: 110 (Connection timed out))
19:35 < thl> | okay, I'll try to do that
19:35 < thl> | for those interested
19:35 < thl> | the list of broken deps is at https://www.redhat.com/archives/fedora-extras-list/2006-March/msg01368.html
19:35 < thl> | warren, can you handle at-poke?
19:35 < scop> | thl, no need to file bugs about any kmod-*
19:36 < thl> | scop, sure :)
19:36 < warren> | thl, yes
19:36 < warren> | thl, I think we need to remove it for now, upstream is just broken
19:36 < thl> | the same holds true for pam_mount
19:36 < warren> | I'll figure something out.
19:36 < warren> | If it is broken, just remove it.
19:36 < thl> | https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=174190
19:37 < scop> | thl, and cernlib and paw will be removed from x86_64/devel, there's a request in the FC5Status page
19:37 < thl> | scop, they were removed already
19:37 < scop> | ah
19:38 < thl> | does anyone know why the *imap* packages are not build?
19:38 < warren> | "imap" itself?
19:38 < thl> | warren, cyrus-imapd-murder cyrus-imapd-nntp up-imapproxy
19:39 < warren> | murder and nntp are not listed in owners
19:39 < thl> | well, I'll try to take care of those
19:39 < thl> | let's proceed
19:40 --- | thl has changed the topic to: FESCo meeting -- Weekly sponsorship nomination
19:40 < thl> | anyone?
19:41 --- | thl has changed the topic to: FESCo meeting -- comps.xml for Extras
19:41 < thl> | jeremy, is that working fine?
19:41 < thl> | can I remove it from the schedule now?
19:41 < jeremy> | thl: seems to be
19:41 < jeremy> | thl: still need to make it automatic, but manual is okay for now
19:41 < thl> | jeremy, okay
19:41 --- | thl has changed the topic to: FESCo meeting -- Extras to handle multilib
19:42 < thl> | anybody interested to drive this forward
19:42 < warren> | I personally don't thiun+
19:42 < warren> | think it is worth it
19:42 < warren> | I mean, for what purpose?
19:42 < thl> | wine.i386 for x86_64?
19:42 < jeremy> | thl: we have to tackle it at some point
19:42 < warren> | oh
19:43 < thl> | jeremy, agreed
19:43 <-- | Sopwith has quit (Connection timed out)
19:43 < warren> | thl, I was thinking more the example given on the list where something was trying to build against both glibc 64bit and 32bit
19:43 < jeremy> | thl: but at the same time, I think trying to before we come up with a good way to do it automatically for core is probably just asking for trouble :-/
19:43 < thl> | warren, that's a core problem mostly ;-)
19:43 < jeremy> | notting and I want to kill the way we're currently doing it for fc6, though, so it may be possible soon
19:43 * | jeremy notes that we want to fix that too ;)
19:44 < thl> | jeremy, okay
19:44 < thl> | but could we find a interrim solution for wine?
19:44 < thl> | simply a small script that checks for a new wine
19:44 < thl> | and copies it over from the i386 repo to the x86_64 repo
19:45 < thl> | that would be a start
19:45 < thl> | it seems many people ask for wine.i386 in the x86_64 repo
19:45 < jeremy> | we could do the same sort of hack we do for core
19:45 < jeremy> | which is a text file listing "multilib packages" and then we depsolve
19:46 < jeremy> | but that'll require some complication to how the extras push stuff works
19:46 < thl> | :-/
19:46 < thl> | would a manual process work for now?
19:46 < thl> | that seems a lot easier
19:46 < thl> | scop ?
19:47 < scop> | the FCxStatus wiki pages would work sufficiently well if it's only wine
19:47 < thl> | k; then I suggest we do it like that
19:47 < thl> | or does anyone dislike the idea?
19:47 < skvidal> | it's more error prone
19:47 < thl> | wine has nearly no i386 deps in extras iirc
19:48 < skvidal> | jeremy: one problem
19:48 < thl> | skvidal, let's try it for now and see how it works
19:48 < thl> | skvidal, what okay?
19:48 < skvidal> | if an i386 pkg in extras requires an i386 pkg from core that's not in the x86_64 tree in core
19:49 < skvidal> | we've got no way to make it available in core
19:49 < thl> | skvidal, I checked that for wine some weeks ago
19:49 < skvidal> | not just talking about wine
19:49 < skvidal> | for any pkg
19:49 < thl> | for any pkg: sure
19:49 < thl> | but the manual process would be only for wine atm
19:50 * | skvidal knows how manual processes tend to get out of hand, quickly.
19:50 < thl> | that would force us to find a proper solution ;-)
19:51 < thl> | okay: manual process for wine: some yes/+1 or no please
19:51 < jeremy> | I'm okay with it for now I guess
19:51 < thl> | I'M okay with it for now, too
19:52 < thl> | scop, ensc, skvidal ?
19:52 < thl> | warren ?
19:52 < skvidal> | I'll live, sure.
19:52 < skvidal> | just wanted to give a bit of a warning :)
19:52 < scop> | sure, still works for me
19:52 < scop> | btw it should be very trivially scriptable too
19:52 < thl> | k, then we do it like that
19:52 < ensc> | I do not like manual ways
19:52 < thl> | scop, feel free to make it scriptable
19:53 < thl> | ensc, the goal is to have a proper solution in hte near future
19:53 < thl> | I think we all agree on that
19:53 < scop> | I'll take a look while tweaking the scripts for post-FE5 devel
19:53 < thl> | scop, thx
19:53 < thl> | k, let's move on
19:53 < ensc> | such temporary hacks tend to become permanent...
19:53 * | skvidal gasps
19:54 < thl> | ensc, doing things somehow sometimes is better then doing nothing
19:54 < skvidal> | I, umm, agree with ensc
19:54 < thl> | ensc, skvidal, okay
19:54 < thl> | can you work out a proper solution in... let's say, three weeks?
19:54 < skvidal> | thl: hold on
19:55 < skvidal> | thl: I didn't say no
19:55 < skvidal> | I just said we need to keep it in mind
19:55 < skvidal> | :)
19:55 <-- | jeremy has quit (Read error: 104 (Connection reset by peer))
19:55 < thl> | :)
19:55 < scop> | what would such a proper solution consist of, apart from making it automatic?
19:55 < skvidal> | thl: so +1 for doing something and +1 for keeping an eye on it
19:55 < warren> | well, jeremy did say he needs to design a proper solution for Core
19:55 < skvidal> | scop: well we need to solve the core/extras problem for i386 pkgs needed from core
19:55 < warren> | we could adopt that when it happens
19:55 --> | jeremy (Jeremy Katz) has joined #fedora-extras
19:55 < scop> | okay
19:55 < thl> | warren, sounds like a plan
19:56 < thl> | okay, so "manual processes for wine and keep an eye on it"
19:56 --- | thl has changed the topic to: FESCo meeting -- Free discussion
19:56 < thl> | anything?
19:56 < skvidal> | I like brocolli
19:57 < warren> | skvidal, me too.
19:57 * | skvidal waits for spot to chime in about pork
19:57 < thl> | just for the record
19:57 < thl> | there are some issues I'd like to solve soon
19:57 < skvidal> | okay
19:57 < thl> | but I'll post to the lists directly about it
19:57 < skvidal> | cool
19:58 --> | adrianr (Adrian Reber) has joined #fedora-extras
19:58 < thl> | I hope to find more time for such stuff when FC5 is out
19:58 < thl> | okay
19:58 * | thl will close the meeting in 30
19:58 < warren> | gotta go to another meeting
19:58 * | thl will close the meeting in 15
19:58 < thl> | warren, bye
19:59 * | thl will close the meeting in 5
19:59 < thl> | MARK meeting end
19:59 < thl> | thanks everyone