15:59:34 <smooge> #startmeeting EPEL meeting for 2010-06-18
15:59:34 <zodbot> Meeting started Fri Jun 18 15:59:34 2010 UTC.  The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:59:34 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:03 <smooge> #topic Roll Call
16:00:07 <smooge> me
16:00:07 <tremble> Here
16:00:10 <derks> present
16:00:14 * Jeff_S 
16:00:41 * skvidal is lurking
16:01:17 <Jeff_S> uhoh
16:01:17 <smooge> #topic Agenda items
16:01:37 <smooge> 1) libtalloc
16:01:42 <smooge> 2) EL-6 readiness
16:02:08 <smooge> 3) new meeting time
16:02:14 <smooge> 4) open floor/other
16:02:26 <smooge> #topic libtalloc
16:02:44 <smooge> anyone want to go over this one?
16:03:00 <smooge> from what I can tell we have something that breaks samba3x
16:03:07 <smooge> and we need to fix it somehow
16:03:18 <tremble> summary: EL included a lib that EPEL's had for a while, but EL included an old version
16:03:35 <smooge> so our fix will be
16:03:35 * skvidal is fine with the suggested policy changes
16:03:44 <skvidal> provided the policy changes do not block FIXING THE PKG
16:03:55 <skvidal> ie: let's not wait on policy approval to get the pkg fixed
16:04:01 * sgallagh is here (late)
16:04:15 <smooge> I would like to get abadger1999 to go over our policy and to clean it up... its a mess
16:04:36 <smooge> 2 I want to get it fixed.
16:05:21 <tremble> I would be tempted to suggest a libtalloc-compat package and smiling sweetly at the RH maitainer to see if he's willing to maintain it until 5.6
16:05:41 <Jeff_S> the proposed solution being creating an updated libtalloc 1.x package to tide us over until EL 5.6?
16:05:49 <smooge> 3) from dealing with various things in EPEL it is clear that our 'policy' is mostly a guideline and should be worded as such. Dealing with the turbogear breakage this week made me realize that we are pushing stuff in that breaks stuff but people only get dinged if they ask first
16:06:39 <sgallagh> I haven't had a chance to prove this theory yet, but I think we can function safely by adding a libtalloc2 package to EPEL and dropping the .so.1 from EPEL. It was only originally added to resolve the naming conflict, but that blew up
16:06:59 <sgallagh> The features in EPEL all rely on the .so.2. Only RHEL 5.5 features need the .so.1
16:07:00 <skvidal> Jeff_S: no the proposed solution is making a libtalloc pkg that has both the 1.X from RHEL5 and the 2.x version in it
16:07:26 <sgallagh> The catch being, of course, that anyone that has ALREADY been upgraded to the EPEL5 version of libtalloc would be stranded
16:07:27 <Jeff_S> if it doesn't break (more) stuff I'm all for it
16:07:31 <skvidal> sgallagh: if you drop the libtalloc 2.0.X from epel then people who have installed it would be stuck
16:07:50 <skvidal> so if you just add the rhel5.5 libtalloc 1.X hacked up version to the libtalloc-2.0.X pkg in epel
16:08:06 <skvidal> then you fix rhel5.5 people using samba3x and having epel enabled
16:08:12 <skvidal> and you don't leave others stranded
16:08:21 * derks thinkgs it would be nice to have a tool that give a good estimate on how many boxes might have xxx package installed
16:08:24 <sgallagh> Which is currently available in https://admin.fedoraproject.org/updates/libtalloc-2.0.1-7.el5
16:08:26 <skvidal> then when rhel 5.6 hits the streets samba3x uses the new one
16:08:33 <skvidal> derks: we have no such data
16:08:38 <skvidal> derks: we cannot collect it
16:09:00 <derks> skvidal, i know...  unfortunately we can't
16:09:01 * abadger1999 here
16:09:23 <tremble> Quite a few places would probably have things tucked away behind private Satellite/Spacewalk servers.
16:09:56 <skvidal> derks: not just that it is outside of the rules to collect it - but it is alos impossible b/c of mirroring
16:10:12 <Jeff_S> stupid mirrors good for nothing
16:10:33 <derks> skvidal... I was thinking more an anonymous tool that reported what packages were installed on a system
16:10:38 <sgallagh> skvidal: Does smolt include installed packages?
16:10:52 <smooge> anyway... lets focus on what we can do :)
16:10:54 <skvidal> yes
16:11:02 <skvidal> sgallagh: I don't remember - but smolt is voluntary
16:11:06 <Jeff_S> do we have a proposed package already?
16:11:09 <abadger1999> yeah, having just a libtalloc2 would not cross any policy but it's really the least friendly of hte options we have for end users.
16:11:13 <smooge> and few enterprise boxes install it
16:11:29 <sgallagh> Jeff_S: See the above package
16:11:36 <smooge> so we would need a libtalloc2 and a libtalloc1
16:11:38 <Jeff_S> sgallagh: thanks, missed that
16:11:47 <derks> smooge, if smolt *is* installed... does it report automatically/frequently?
16:12:04 <sgallagh> It's carrying libtalloc from samba3x (rebuilt, not binary) as well as libtalloc 2 from upstream.
16:12:07 <smooge> derks, if it is installed AND configured AND allowed to talk out yes.
16:12:13 <abadger1999> I like us having both libtalloc-2 and libtalloc1-1 packages -- skvidal would rather have a single libtalloc package that has both libtalloc1 and libtalloc2 binaries in it.
16:12:43 <skvidal> I think splitting it means more maintenance work for a temporary item
16:12:46 <skvidal> but really
16:12:59 <sgallagh> abadger1999: You're suggesting libtalloc1 Obsoletes: libtalloc, right?
16:12:59 <skvidal> I'm not that strongly for it - I just want it fixed and I don't  want to wait on policy to get it fixed
16:13:09 <abadger1999> Both of those need us to amend policy but one is epel-only policy and the other is both epel policy and packaging guidelines.
16:13:20 <sgallagh> That would then break people who need libtalloc 2 who already have it installed, yes?
16:14:49 <abadger1999> i don't think so but skvidal will know better.
16:14:58 <derks> Wouldn't it be possible to create libtalloc1 (obsoletes libtalloc < 2) and libtalloc2 (obsoletes libtalloc >= 2)?
16:15:17 <skvidal> abadger1999: so the problem is anytime we involve an obsoletes we walk down a path that's harder to come back from
16:15:18 <sgallagh> derks: No, because that still leaves people who need BOTH stranded
16:15:18 <abadger1999> The packages should all have dependencies on the library sonames so they should still pull in the correct packages.
16:15:40 <skvidal> abadger1999: also remember that rhel5's yum is not as fancy-pants as f13s in some ways
16:15:58 <skvidal> abadger1999: so some of compare_providers is not the same
16:16:05 <abadger1999> <nod>
16:16:22 <derks> sgallagh, gotcha.  What if libtalloc1 also requires libtalloc2? (dirty, but works)
16:16:43 * sgallagh twitches
16:16:52 * derks not sure it would make any sense to do it that way
16:17:02 <tremble> Also note that when EL5 adds libtalloc we'd have already obsolleted it.
16:17:14 <sgallagh> gah, that's also true
16:17:23 <smooge> ugh ok lets go with skvidal's one
16:17:31 <smooge> and fix policy
16:17:38 <smooge> s/policy/guidelines/
16:18:19 <sgallagh> smooge: Policy should also address the related case where EPEL is using a newer version but NOT an soname bump
16:18:20 <smooge> proposal: have libtalloc include both libtalloc1 and libtalloc2 until RH releases 5.6 next ice age to fix current issue
16:18:52 <sgallagh> e.g. libsomething 1.1.0 vs libsomething 1.0.0 is backwards compatible, but they both have .so.1 and RHEL decides to only ship 1.0.0 after EPEL
16:19:10 <abadger1999> So I sent wording for how to show that EPEL allows bundling of libraries in certain cirumstances to the lis, if you accept that here, then it takes care of the packaging guideline conflicts.
16:20:02 <abadger1999> Someone should iron out all the details of the wordingfor the EPEL policy piece (No conflicts with RHEL packages)
16:20:54 <abadger1999> The EPEL policy piece can also include how to deal with incompatible version with no soname bump.  (rpath might work)
16:21:08 <sgallagh> I thought rpath was explicitly forbidden
16:21:12 <abadger1999> Or just let them conflict.
16:21:31 <abadger1999> Well, bundled libraries are too --  epel can override packaging guidelines if it chooses.
16:21:41 <abadger1999> Also -- rpath is not forbidden for private libraries.
16:21:46 <sgallagh> ok
16:22:00 <abadger1999> But the definition of "private libraries" is being stretched a bit here.
16:22:24 <abadger1999> Would be good for EPEL to make clear what the intention is.
16:22:45 <smooge> ok sgallagh are you adverse to bundling both for a while so that we don't have broken people?
16:23:03 <sgallagh> smooge: A package to do exactly that is already in epel-testing
16:23:06 <abadger1999> https://fedoraproject.org/wiki/Packaging:Guidelines#Rpath_for_Internal_Libraries
16:23:23 <sgallagh> That was my original idea, but I wanted to make sure it got buy-in from the Powers-That-Be before it went stable
16:24:07 <smooge> abadger1999, I want to go over our guidelines as a do-over.. they read like ancient hebrew mixed with cyrillic to me
16:24:24 <smooge> abadger1999, but I would like to go over that with you after meeting
16:24:31 <abadger1999> smooge: Okay.
16:25:18 <smooge> ok anyone have objections to what is in epel-testing currently? And could we get some testing of it?
16:26:11 <smooge> I will take the silence as "ok"
16:26:52 <smooge> #topic EL-6 readiness
16:27:10 <tremble> Many packages still not tagged and built
16:27:30 <sgallagh> smooge: Yeah, I'd like to see more testing of that package. Not sure the right approach to fixing it.
16:27:42 <smooge> tremble could I ask a favor. Could you make a list?
16:27:51 <sgallagh> s/fixing/advertising/
16:28:28 <tremble> smooge: 5489 pkgs vs 1138  very long list
16:28:48 <sgallagh> This should probably see an announcement on the epel-devel mailing list
16:29:01 <smooge> sgallagh, post to the list and a test plan to see how one could 'break' their system and how this would 'fix' it
16:29:02 <sgallagh> I for one assumed a mass rebuild was going to happen with packages split from EPEL5
16:29:13 <sgallagh> smooge: Ok, I will do so
16:29:44 <smooge> sgallagh, no we weren't doing a mass rebuild but everyone assumes we would
16:29:56 <smooge> I think we are just going to have to do so in July
16:30:01 <sgallagh> smooge: Yeah, hence my suggestion about making an announcement
16:30:20 <smooge> sgallagh, oh I thought you meant an announcement about libtalloc
16:30:40 <sgallagh> smooge: Sorry, stupid antecedents keep running away from me
16:30:45 <tremble> On the perl side I'm finding a number of places where deps were gained between F6 and F12
16:32:15 <smooge> tremble, I am an idiot.. not sure how one goes about building it as I would like to get it... if I can get it I will put it up as a page on fedorapeople to point people to
16:32:25 <tremble> On which note, does anyone know if cweyl is interested in EPEL (he's not listed but I got a go ahead and branch last time I asked about a package)
16:32:46 <smooge> to be honest I think the interested page is probably a dead tree at this point
16:33:40 <smooge> its one of those things I need to talk with abadger1999 about as its policy but no one follows/uses it
16:33:50 <tremble> Should I just ask for an EL6 branch on the strength of his previous IRC comment?
16:33:54 <smooge> yes
16:34:00 <tremble> Ok
16:34:09 * tremble follows it.
16:34:26 <smooge> i think one of our policy changes will be dealign with asking/getting branch changes
16:34:56 <smooge> I know some people follow it and others well ... better to do and ask forgiveness later or some such
16:36:23 <smooge> ok anything else on EL-6?
16:36:30 <tremble> Would be nice not to need to spend 2 weeks waiting for a non-responsive maintainer.
16:37:09 <skvidal> tremble: but it sucks to come back from vacation and find out you got your packages took
16:37:10 <skvidal> :)
16:37:21 * stahnma is here now
16:37:23 <stahnma> :)
16:37:58 <sgallagh> skvidal: Well, can we establish a better vacation policy?
16:38:11 <tremble> skvidal: true, however I doubt someone trying to get a batch of packages into EPEL would bitch if asked to surrender control back to their Fedora maintainers.
16:38:11 <skvidal> sgallagh: I'd like more vacation time :)
16:38:32 <skvidal> maybe - depends on how much hatefulness there is :)
16:38:43 <skvidal> it's not like we've never seen animosity between packagers :)
16:38:52 * skvidal rolls his eyes with sarcasm
16:39:00 <tremble> skvidal : Ergh
16:39:33 <tremble> smooge: do you want me to see if I can build that list (would probably be some point next week now)?
16:39:55 <smooge> yes
16:39:59 <smooge> tremble, yes please
16:40:17 <sgallagh> tremble: Quick question about EPEL6 builds. Is building enough, or do we need to go through updates process as well?
16:40:23 <sgallagh> (Given that EPEL6 is still a beta)
16:40:40 <tremble> We're just building straight into EPEL6 at the minute
16:40:47 <sgallagh> Ok, thanks.
16:41:00 <sgallagh> So all the packages I just built will appear automagically, then :)
16:41:16 <tremble> Once kojira kicks in, yes
16:41:38 <tremble> As will the 20+ packages I just threw through.
16:41:57 <smooge> is trying to get whenisgood to be obvious
16:41:58 <smooge> one sec
16:42:52 <smooge> #topic Possible new meeting time
16:43:09 <smooge> sgallagh, yes
16:43:39 <smooge> ok I was confused when looking at the whenisgood and seeing all these times with 7 on them. I thought that meant all 7 people could make it
16:43:51 <smooge> but it turns out that no it means none could make that time
16:44:19 <smooge> Anyway... the best time looks like Monday at 1pm
16:44:27 <smooge> ugh sorry that would be 1pm my time
16:44:46 <smooge> 1900 UTC
16:44:56 * tremble nods
16:45:41 <smooge> I will post these to the list in a short while... but I think we will do next week at THIS time and then move to #fedora-meeting-1 or whatever on the new time.
16:45:53 <smooge> does that sound good.
16:46:25 <smooge> crickets
16:46:44 <tremble> Doesn't sound bad
16:46:52 <smooge> ok last topic then close
16:46:56 <smooge> #topic Open Floor
16:47:10 <smooge> bueller? anyone seen bueller?
16:48:25 <sgallagh> smooge: How is the EPEL steering committee elected?
16:49:07 * sgallagh realizes that this sounds like a loaded question
16:49:15 <smooge> we don't. its appointment for life
16:49:40 * tremble assumed it was mostly based off decisions on here and the list.
16:49:42 <smooge> sgallagh, we had elections but found that as soon as someone was elected life would get in their way and take away anytime they had for EPEL
16:50:18 <sgallagh> So then there's no opportunity for new individuals to step up?
16:50:27 <smooge> Oh everyone can step up
16:51:20 <smooge> sorry the appointment for life was meant to be a joke
16:51:58 <smooge> I am not really sure there is a "Steering Committee" anymore. The last 2 really had the issue of all of us having big issues come up as soon as we were elected to it
16:52:14 <smooge> its another thing I need to work with people on to clean up and deal with
16:52:15 <sgallagh> So has it just been rolled into FESCo then?
16:52:31 <smooge> I am not sure FESCO wants us :P
16:53:10 <tremble> Are we any different to any of the other SIGs? (except we get whole repositories to play with)
16:53:17 <smooge> sgallagh, you are raising an important question.
16:53:52 <smooge> well at one point we werent a SIG but something else  as we didn't fit exactly into SIG roles
16:54:56 <abadger1999> epel is bigger than a sig
16:54:59 <smooge> sgallagh, could I ask a favor and you bring it up on list as we are hitting the end of the meeting... I think we need to focus on our emperor's clothing
16:55:03 <abadger1999> And kinda to the side
16:55:34 <sgallagh> smooge: Ok, I'll raise the topic on the list.
16:55:38 <smooge> we are sort of the horses rear end (as our logo seems to look as some people)
16:55:42 <abadger1999> ie: epel and fedora could exist independently of each other except that we share the infrastructure/it resources.
16:56:53 <Jeff_S> yes, there is a small group that tend to show up and do stuff for EPEL, but it lost the feeling of an "official" steering committee some time ago
16:56:58 <smooge> abadger1999, I agree. we really need some better governence because of that. HOWEVER we are also about 5 interested people and 95 people who just like make packages
16:57:20 <smooge> well 7 people since that is how many signed up for meetings
16:57:45 <tremble> Might be 6, I was lazy the second time.
16:58:53 <smooge> ok I need to close the meeting for the next crew
16:59:04 * sgallagh goes to lunch
16:59:08 <smooge> mailing list for the rest
16:59:11 <smooge> see you guys later
16:59:14 <smooge> #endmeeting