19:30:48 <tremble> #startmeeting EPEL (2010-10-11)
19:30:48 <zodbot> Meeting started Mon Oct 11 19:30:48 2010 UTC.  The chair is tremble. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:30:48 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
19:31:00 <tremble> #meetingname epel
19:31:00 <zodbot> The meeting name has been set to 'epel'
19:31:09 <tremble> #chair nirik smooge
19:31:09 <zodbot> Current chairs: nirik smooge tremble
19:31:19 <tremble> #topic init process / agenda
19:31:27 <tremble> So current items...
19:31:27 * nirik waves
19:31:31 * stahnma hi
19:31:39 * tremble is here
19:31:49 * derks is here
19:32:07 <smooge> here
19:33:16 <tremble> * EPEL Bug numbers
19:33:16 <tremble> * Broken dependencies in stable (stahnma)
19:33:16 <tremble> * Conflicting Packages Policy  (derks)
19:33:16 <tremble> * Rubygem-rack Version
19:33:16 <tremble> * Open floor
19:33:33 <tremble> Anyone got anything else to add to the list before we begin?
19:33:34 <nirik> sounds good. I have a update on lua as well if anyone cares.
19:33:44 <stahnma> I imagine somebody cares
19:33:47 <tremble> * lua (nirik)
19:34:36 <tremble> #topic EPEL Bug List
19:34:48 <nirik> still under 200. ;)
19:35:04 <tremble> Unfortunately they're going up again at the minute :(
19:35:24 * nirik notes 6 of them are clamav...
19:35:27 <derks> nirik, does that include package reviews?
19:35:34 <tremble> derks: No
19:35:36 <nirik> derks: nope.
19:35:37 <stahnma> very package reviews specifically for epel
19:35:40 <smooge> no ..
19:35:41 <stahnma> if any...
19:35:41 <derks> eesh
19:35:47 <smooge> hey. I have some
19:35:56 <tremble> stahnma: There's a few parallel install ones
19:35:59 <stahnma> ok, back to my very few tatement :)
19:36:04 <stahnma> +s
19:36:34 <tremble> At some point we need to get around to reviewing smooges packages since they kill off a few bugs.
19:36:45 <stahnma> true
19:36:58 <stahnma> I have a bug that will close with a stable push in a few days
19:37:06 <nirik> I could try to do some reviews in the upcoming week.
19:37:08 <stahnma> and probably a few others I need to update/catch up on
19:37:41 <tremble> I've a couple of bugs waiting on time in testing which should close in the next week or so.
19:37:48 <nirik> also, I see 2 bugs that are filed against a busy coworker here, and I will see if he's ok with me just fixing them.
19:38:34 <tremble> Some of them are just so old I wonder if people have forgotten to check for them
19:38:39 <nirik> hopefully the lua ones will be closed before too long too.
19:38:53 <tremble> A few saw motion just with the Whiteboarding prompting people.
19:39:06 <stahnma> sometimes that's what it takes
19:39:08 <nirik> we agreed that things dropped in beta2 can just be branched normally now right?
19:39:08 <stahnma> just a reminder
19:39:16 <tremble> nirik: yes
19:39:18 <stahnma> I think we did
19:39:27 <nirik> so the perl-TK bug could be dealt with that way.
19:39:48 <tremble> Yes, hence the branch request that went in just after your last run :)
19:40:31 <nirik> yep.
19:41:06 <tremble> There's another half dozen or so that're just going to have to wait for 6-GA when we know what shape the channels are going to be in.
19:41:29 <tremble> But can then be closed off fairly quickly
19:41:39 <nirik> yeah.
19:41:40 * stahnma wonders if we should try to triage/fix/close el4 nowish, as many may have little/no action/solution
19:42:07 <derks> tremble, any info on when 6-GA will drop?
19:42:17 <tremble> derks: Nothing I can talk about.
19:42:27 <derks> right on
19:42:36 * schlobinux_ handles a beer to tremble :-)
19:42:50 <nirik> 15 bugs are el4
19:42:50 * tremble already has a beer :)
19:42:58 <stahnma> doing it right
19:43:27 <tremble> I think we WONTFIX old bugs on EL4 when 6 hits GA
19:43:50 <tremble> Unless the owner's making some form of movement...
19:44:08 <tremble> Treat it like the Fedora cycles.
19:44:31 * SmootherFrOgZ just get in
19:44:31 <nirik> well, except el4 is not end of life at all.
19:44:49 * nirik doesn't see any reason to wontfix them until support ends...
19:45:15 <smooge> I think that we should keep fixing on EL4
19:45:21 <tremble> Ok
19:45:23 <stahnma> I agree
19:45:41 <smooge> even with 4.9 out some day A LOT OF PEOPLE use 4
19:45:46 <derks> unfortunately I'd have to agree with nirik ... being in an environment where customers are pissed off about el3 being eol... I think it'd be best to continue to fix until end of support
19:45:47 <stahnma> although, RH won't fix things on EL4 unless its security
19:45:47 <smooge> more than I would guess 6 :)
19:46:11 <smooge> hehehe. I got asked if I could fix EL-3 for some people on the side.
19:46:12 <nirik> yeah, might be good to sometime poke those 15 bugs and see which are just not fesable to fix tho.
19:46:14 <stahnma> perhaps we should marry against the RHEL support lifecycle
19:46:38 <tremble> stahnma: And WONTFIX the non-sec ones?
19:46:52 <tremble> or at least the RFEs
19:47:25 <derks> tremble, yeah... RFE's should be a no-go for epel-4 in my opinion
19:47:31 <stahnma> https://access.redhat.com/support/policy/updates/errata/
19:47:44 <stahnma> maybe we should do some homework and discuss next meeting
19:48:20 <nirik> sounds reasonable.
19:48:54 * tremble nods
19:48:55 <stahnma> #info  https://access.redhat.com/support/policy/updates/errata/ RHEL Lifecycle overview
19:49:06 <stahnma> nod sure if my info worked
19:49:12 <tremble> stahnma: Fancy trying to put together a draft proposal ?
19:49:17 <tremble> That will have worked.
19:49:27 <derks> would it be helpful at all to ping mdomsch and see how many uniques are checking in for epel-4 on mirrormanager... just to have an idea of usage?
19:49:27 <tremble> It just doesn't give any output
19:49:44 <stahnma> tremble: I'll see if I get some time for that
19:49:55 <stahnma> tremble: I know the rhel support lifecycle pretty well, so probably
19:50:37 <tremble> stahnma: Well there's no major rush about it yet.
19:51:02 <tremble> derks: To be fair I know of far too many places still making heavy use of 4
19:51:27 * stahnma too
19:51:36 <stahnma> I just don't think we can do a ton with epel for it
19:51:46 <tremble> derks: I suspect most places with a significant number of hosts will hide them behind private mirrors
19:51:53 <stahnma> specifically if you are on RHEL because up2date can't handle proxys and such
19:51:55 <derks> tremble, same here ... just curious of epel-4 ... personally we don't see a lot of usage of epel on rhel 4... but we have a ton of rhel 4 boxes
19:52:26 <nirik> possibly in part due to epel coming out after el4 was released... so many people rolled their own stuff or made due without.
19:53:29 <mdomsch> I dont' have any numbers off hand
19:53:35 <stahnma> ok
19:53:43 <mdomsch> we could generate them I suspect
19:53:55 <stahnma> should  we move on?  Next week we can continue the discussion of support for EL4 after doing our homework
19:54:10 <derks> mdomsch, I was just throwing out the idea... i don't really think it would affect our decisions here
19:54:22 * nirik votes move on.
19:54:30 <tremble> #topic Broken dependencies in stable (stahnma)
19:54:34 <stahnma> ++
19:55:39 <tremble> #info EL4 Numbers curtsey of dgilmore ... 1518 for i386 and 886 for x86_64
19:55:46 <nirik> over what time?
19:56:16 <tremble> #info el5 has 87023 x86_64 and 78071 i386, and 81 ppc
19:57:13 <dgilmore> #info over 7 days
19:57:24 <tremble> stahnma: want to take the lead on this topic?
19:58:34 <smooge> yeah broken deps
19:59:48 <stahnma> just a sec
19:59:49 <stahnma> brb
19:59:52 <stahnma> dayjob
20:00:09 <tremble> So we agreed last time we'd unpush the el5 (and el4?) STABLE packages with broken deps that weren't going to be easy to fix.  Having already tidied the ones which just needed a rebuild...
20:00:48 <stahnma> ok back
20:00:49 <nirik> tremble: yeah.
20:00:50 <stahnma> sorry
20:01:00 <nirik> do we have a list? shall we just go over them one by one? or ?
20:01:00 <stahnma> tremble: yes
20:01:05 <stahnma> I sent a pastie
20:01:14 <stahnma> http://fpaste.org/ctkI/
20:01:19 <stahnma> tremble did some updates
20:01:22 * nirik did a push this morning, worth re-running?
20:01:30 <stahnma> I was waiting for mirrors to populate
20:01:46 <stahnma> which seems to take a while
20:01:47 <stahnma> i can try
20:02:11 <tremble> It's a lot better than it was, stahnma did a fair bit of work on the rubygem trees
20:02:17 <nirik> ok. yeah. Thanks stahnma
20:02:28 <nirik> I'd be happy to unpush things later today.
20:02:35 <stahnma> there are two ruby packages we need to unpush
20:02:43 <stahnma> and the rest can be fixed
20:03:02 <stahnma> and one might be able to be fixed, but will require a parallel install
20:03:09 <stahnma> pre-req
20:03:23 <stahnma> I haven't looked at 4 closely enough
20:04:02 <tremble> probably easier to unpush them, then push them back out with their deps.
20:04:18 <tremble> if that means they need a bump, then oh well.
20:04:33 <nirik> yeah, and many of these have been around for ages.
20:04:33 <stahnma> ++
20:04:53 <stahnma> yeah, the only things I really don't want pulled are rubygem-sinatra and rubygem-shotgun
20:04:57 <nirik> I see 22 that have no comment or a note that they can't easily be fixed.
20:04:57 <stahnma> those are fixable still
20:05:07 <stahnma> yeah, the no comment ones I didn't get time to dig into
20:05:14 <stahnma> the ruby stack has been enough fun :)
20:05:40 <stahnma> rhnlib and xulrunner-devel should be left alone due to centos/el differences
20:06:31 <nirik> rhnpush you mean?
20:06:39 <stahnma> possibly :)
20:06:47 <stahnma> yes
20:06:50 <stahnma> the dep on rhnlib
20:06:52 <stahnma> :)
20:07:19 * stahnma will turn in a rel-eng ticket for unpushes for EL5
20:07:30 <stahnma> should we go ahead and do everything broken in EL4 as well?
20:08:18 <nirik> http://fpaste.org/T4Mq/
20:08:47 <nirik> I'd say lets perhaps look at el4 next week?
20:08:59 * tremble nods
20:09:03 <stahnma> ok
20:09:07 <stahnma> +1 for that
20:09:12 <tremble> unpush fwsnort
20:09:37 <tremble> I'll repush it with a dep fix and the dependency once I've built it.
20:09:41 <nirik> ok.
20:09:58 <tremble> They're doing a hard dep on a perl package (rather than the module)
20:10:14 <nirik> http://fpaste.org/ktpx/
20:10:42 <tremble> +1
20:11:00 <nirik> so, we want to look at the possibly ones more? or just do them too?
20:11:37 <stahnma> I'd vote to unpush the majority of them, but I don't know what they do
20:12:02 <tremble> I vote for unpush, on the grounds that they're not installable.
20:12:08 <stahnma> +1
20:12:16 <nirik> ok. Sounds reasonable to me.
20:12:21 <stahnma> if they get fixed, we put them back
20:12:24 <stahnma> it's that simple
20:12:29 * nirik can make with the clicking later today.
20:12:54 <stahnma> ok
20:12:54 <gomix> !
20:13:00 <stahnma> next item?
20:13:36 <tremble> #topic Conflicting Packages Policy (derks)
20:13:36 <nirik> gomix: you had something?
20:13:50 <gomix> just heard something about fwsnort... and the pacakger
20:13:56 <tremble> ?
20:13:59 <gomix> did something wrong?
20:14:26 <gomix> i mean im the packager
20:14:28 <tremble> There's a package it depends on that's not in EL5
20:14:36 <tremble> Well it will be in 2 weeks
20:15:24 <gomix> Ok, i did not noticed that
20:15:28 <tremble> you're also depending on the package rather than the perl(Module::Name) which is a minor faux pax
20:15:37 <tremble> pas even
20:15:43 <gomix> and builded it over koji
20:16:00 <tremble> Yeah it's runtime rather than build time
20:16:03 <tremble> They happen.
20:16:13 <tremble> It's one of the things we're trying to clean up now.
20:16:18 <gomix> kk
20:17:20 * gomix taking notes... enough for me...
20:17:27 <tremble> nirik: Probably worth emailing $PKG-owner with the reason the package got unpushed too?
20:17:45 <nirik> tremble: I am adding comments to the updates when unpushing.
20:17:51 * tremble nods
20:17:52 <stahnma> as far as conflicting package policy, I like the spirit of having them, as it allows more flexibility.  I'm afraid it might be introducing a maint nightmare
20:18:18 <tremble> derks: You wanted this brought up, anything specific you wanted to say?
20:18:25 <stahnma> we have very few maintainers who are really keeping up with everything in EPEL as it stands today.  Allowing differences from the Fedora packaging guidelines might be difficult.
20:18:49 <derks> I really just wanted to get a definite decision on how to handle it
20:19:08 <derks> stahnma, so you are saying 'no' to the hard 'Conflicts' yeah?
20:19:16 <stahnma> derks: no, I am not
20:19:18 <tremble> Was there a specific example ?
20:19:30 <stahnma> derks: I am saying it takes extra care
20:19:32 <derks> tremble, yes... python26-mod_python, and python26-mod_wsgi
20:19:40 <stahnma> and I'm not certain our maintainers currently have it
20:19:56 <derks> i've submitted reviews for both this week
20:20:08 <tremble> Ah so we're talking about the parallel install packages that aren't going to be around for fedora anyway...
20:20:17 <derks> tremble, exactly
20:20:22 <stahnma> yeah, it makes sense
20:20:27 <stahnma> and I know people are doing it in-house anyway
20:20:42 <schlobinux_> just a note this is not installable in parallel
20:20:54 <schlobinux_> well, not all of the IUS packages I mean
20:20:59 <derks> to me, it is confusing to have a package installable... that won't function unless you know what your doing... rather than having a package *not* install because it conflicts with something else... but that is my opinion
20:21:03 <stahnma> ah
20:21:24 <stahnma> derks: I agree for the most part
20:21:34 <derks> schlobinux_, I know... currently I am using IfModule to check for the apache module and only load if it isn't loaded
20:21:39 <stahnma> I am not strongly opinionated on this either way.  I just want quality packages ;)
20:21:42 <tremble> Although in theory you could be running two apache instances...
20:22:14 <derks> tremble, that is a good point as well... in which case you could have both packages installed and not conflict
20:22:17 <nirik> do we want to ask the FPC their thoughts on this?
20:22:19 <smooge> yeah at which point you are not running from default
20:22:26 <nirik> I know toshio chimed in a bit.
20:22:49 <tremble> nirik: FPC at the very least probably have extra thoughts as to the pros and cons.
20:23:04 <derks> toshio mentioned that fedora has a no conflict policy, but that the epel use case might call for explicit conflicts
20:23:18 <schlobinux_> derks: my understanding is python26 is the first step, there is a lot more eventually coming in, like php52, php53, mysql51, etc..
20:23:49 <schlobinux_> and this is going to be really useful, especially if RHEL6 release cycle is aslong as RHEL5
20:23:52 <tremble> schlobinux_: Only if people have the time/inclination to maintain the extra package trees.
20:23:57 <derks> schlobinux_, I'm not aware of all that... I run the IUS community project and we have talked with EPEL (here) about merging efforts... and it always comes down to keeping them separate
20:24:20 <tremble> derks: Which was the IUS project?
20:24:38 <derks> tremble, iuscommunity.org ... dl.iuscommunity.org
20:24:49 <derks> packages that explicitly replace those in rhel
20:24:55 <derks> php52, php53, mysql50, mysql51, etc
20:25:06 <derks> in IUS... we rely on hard conflicts
20:25:20 <dgilmore> derks: things like php versions etc should eb paralell installable
20:25:21 <stahnma> I think many customers/users would love that
20:25:24 <stahnma> if it was available
20:25:39 <stahnma> parallel would be ideal..
20:25:48 <nirik> sure, but which one is used by default?
20:25:56 <derks> dgilmore, it could be...  though some times requires a lot more maintainence over head/patching/etc
20:26:03 <dgilmore> os default or setup alternatives
20:26:20 <nchauvet> ? with parallele doesn't that mean both at the same time ?
20:26:28 <derks> nchauvet, yes
20:26:30 <nirik> yep
20:26:34 <tremble> dgilmore: Except you need to produce alternatives to Red Hat maintained packages.
20:26:48 <tremble> nchauvet: Where possible.
20:27:01 <nirik> so if you run a 'yum install python26-mod_python' it doesn't work until you manually tweak configs. which is kinda confusing.
20:27:06 <derks> the route that ius has taken, is similar to how redhat proper does... if anyone has looked at gcc44 and postgresql84
20:27:24 <derks> gcc44 is parallel... postgresql84 conflicts with postgresql, and provides it
20:27:29 <nchauvet> this has produced weird results with dependencies on debian, specially with 'underlinked'  librarie
20:27:31 <derks> to it replaces postgresql
20:27:31 <dgilmore> tremble: its possible to do
20:27:42 <nchauvet> doesn't it
20:28:08 <tremble> dgilmore: True, would need cooperation from the RHEL maintainers though.
20:28:09 <nchauvet> doesn't fedora has a guideline to prevent this most of the time ?
20:28:28 <nirik> nchauvet: yes, conflicts are frowned upon.
20:28:32 <derks> nirik, the way python26-mod_python works now is... it *will* work out of the box... unless mod_python is installed, then it just silently doesn't load .. which is... confusing
20:28:33 <tremble> nchauvet: yes, but EPEL is different due to the massively longer support cycle.
20:29:37 <nirik> could someone generate a pros/cons and possible guidelines for this? I'm afraid I don't have the brains right at the moment to know where I stand on it. ;)
20:30:20 <tremble> derks: Would you feel up to trying to do that?
20:30:59 <derks> tremble, I can.  I more or less put all of my talking points in the thread on epel-devel
20:32:09 <tremble> #action derks Get the conversation started on epel-devel...
20:32:09 * nirik re-reads
20:32:55 <tremble> Since it's getting late let's move on...
20:32:56 <stahnma> $dayjob meeting
20:32:59 <stahnma> gotta go
20:33:02 <stahnma> byer
20:33:09 <nirik> bye stahnma
20:33:25 <tremble> #topic Rubygem-rack Version
20:34:20 * derks stepping away for a minute
20:34:23 <tremble> stahnma wanted to talk about this, I can't remember the pros and cons, so unless there are objections shall we delay until next week and move on to open floor?
20:34:31 <nirik> yeah
20:35:09 <tremble> #topic Open floor
20:35:43 <nirik> just FYI, I have unpushed all the for sure unpush ones. will work through the rest as time permits.
20:35:53 * tremble thanks nirik
20:38:04 * tremble will close the meeting in 1 minute
20:38:38 <tremble> Since we're already running late and no-one as piped up.
20:39:31 <tremble> #endmeeting