19:00:01 <nirik> #startmeeting EPEL
19:00:01 <zodbot> Meeting started Wed Jan 20 19:00:01 2016 UTC.  The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:01 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
19:00:01 <zodbot> The meeting name has been set to 'epel'
19:00:02 <nirik> #meetingname EPEL
19:00:02 <nirik> #chair smooge avij bstinson dgilmore nirik Evolution
19:00:02 <nirik> #topic Init process / agenda gathering
19:00:02 <zodbot> The meeting name has been set to 'epel'
19:00:02 <zodbot> Current chairs: Evolution avij bstinson dgilmore nirik smooge
19:00:11 <smooge> ah man I had that queued up
19:00:14 <smooge> thanks nirik
19:00:14 <nirik> hey all, who's around for an epel meeting.
19:00:17 <nirik> oh, sorry smooge.
19:00:26 <orionp> hello
19:00:29 <smooge> no problem
19:00:30 <bstinson> hi all
19:00:31 <nirik> tibbs: ^ in case you got sidetracked
19:00:38 <Evolution> morning
19:01:00 <nirik> anyone have agenda items?
19:01:07 <smooge> I have 2 items
19:01:15 <smooge> 1. tibbs work on epel-6
19:01:23 <tibbs|w> Hey.
19:01:44 <smooge> 2. FOSDEM open table on Jan 31
19:01:51 <nirik> cool.
19:02:04 <nirik> I'd like to touch on python3 in epel7.
19:02:09 <smooge> cool
19:02:12 <nirik> anyone have other items?
19:02:25 <orionp> Alternate install prefixes
19:02:26 <Evolution> additional arch support?
19:02:33 <orionp> New package notification for Fedora EPEL
19:03:07 <nirik> cool. ;) thats a nice pile, shall we see if we can get through them?
19:03:07 <smooge> Evolution, oooh that was my number 3. I missed that
19:03:39 <nirik> #topic epel6 macros / rebuild - tibbs
19:03:56 <nirik> tibbs|w: want to talk about this? I think we mostly know where it's at, but would be good for the record.
19:04:01 <tibbs|w> Well, hopefully I'm done with stage 1.
19:04:18 <tibbs|w> Which is getting _something_ in, and being able to verify that it doesn't change anything in existing packages.
19:04:19 <bstinson> tibbs|w: fantastic work so far
19:04:29 <nirik> yep. Good stuff.
19:04:44 <tibbs|w> I've filed the releng ticket so hopefully that will get into the buildroot, and then I can update some packaging guidelines.
19:05:04 <tibbs|w> There's a TODO list at https://pkgs.fedoraproject.org/cgit/rpms/epel-rpm-macros.git/tree/TODO?h=el6
19:05:30 <tibbs|w> I will branch off an EPEL5 package soon and see what I can do there, though I see EL5 stuff as a bit less important these days.
19:05:57 <tibbs|w> If anyone has suggestions for other macros which would make things more convenient/less ifdef-y, please let me know.
19:06:20 <tibbs|w> Technically the same goes for epel7, though there are far fewer differences there.
19:06:30 <nirik> there's a request for mono stuff, but I guess thats epel7...
19:06:32 <nirik> https://bugzilla.redhat.com/show_bug.cgi?id=1295117
19:06:35 <tibbs|w> As an aside, a lot of that EPEL6 stuff is... crufty.
19:06:45 <tibbs|w> Mono should be pretty trivial.
19:07:05 <nirik> yep. crufty indeed.
19:07:14 <nirik> ok, anything else on this?
19:07:35 <tibbs|w> Basically, I don't like complicated packaging.  I don't particularly care about EPEL6, but anything I can do in order to keep stuff cleaner in newer releases is stuff I want to do.
19:07:56 * nirik nods. having the macros will be nice.
19:08:13 <tibbs|w> Anyway, %license is there as soon as this gets in the buildroot.
19:08:42 <nirik> cool.
19:08:57 <nirik> I expect the epel5 stuff will be even more crufty.
19:09:16 <nirik> the vast majority of our users are on 6 though.
19:09:17 <smooge> epel5 mass rebuild :)
19:09:38 <smooge> well 60%.. 35% epel5 and 5% epel7
19:09:41 <tibbs|w> If mock on F23 will actually build EL5 stuff, sure, I'll do a rebuild.
19:09:51 <nirik> it should
19:10:05 <tibbs|w> Hopefully other tools like rpmdiff will still work.
19:10:24 <nirik> they should. if not, it's a bug.
19:10:27 <tibbs|w> I can at least build EPEL6 in about an hour now.
19:10:42 <nirik> I think epel6 has the most packages too...
19:10:44 <smooge> tibbs|w a rebuild of EPEL5 would probably require some mylanta and a bottle of gut rock scotch :)
19:10:49 <nirik> 5 and 7 would be much smaller.
19:10:50 <smooge> nirik, it does.
19:11:06 <tibbs|w> Working on adding another 150 or so cores to my build pile but I don't think it will speed anything up.
19:11:27 <nirik> ok, shall we move on then?
19:11:34 * orionp was hoping for koschei for epel...
19:11:47 <smooge> orionp, its call tibbs|w house
19:11:51 <smooge> nirik, yes..
19:11:57 <tibbs|w> I can do occasional mass builds if people would like those.
19:11:57 <nirik> orionp: we could ask for it... not sure what would be needed for it.
19:12:11 <tibbs|w> Once a week is probably too soon, though.
19:12:30 <orionp> nirik: I asked I think, but was shot down, at least for the moment
19:12:31 <nirik> yeah, once a quarter or something might be nice.
19:12:41 <nirik> orionp: huh, wonder why. ok.
19:12:43 <tibbs|w> But I think koschei and mass builds are kind of a different thing, I think.
19:13:04 <orionp> something about stable releases being different somehow
19:13:06 <nirik> well, koschei will tell you when some change breaks other things... but things change pretty slowly in epel.
19:13:10 <tibbs|w> Maybe I can occasionally post a diff: X started failing, Y now builds.
19:13:18 <tibbs|w> Just have to do more scripting.
19:13:19 <nirik> or perhaps every rhel point release.
19:13:34 <tibbs|w> And I'm also seeing a number of "sometimes it builds, sometimes it doesn't".
19:13:39 <nirik> since the changes happen there mostly.
19:13:45 <tibbs|w> But that's a different topic, I guess.
19:13:49 <nirik> yeah.
19:13:55 <nirik> #topic FOSDEM - smooge
19:14:04 <nirik> smooge: whats going on at FOSDEM? :)
19:14:26 <smooge> ok so Karsten and I will be hosting a round table discussion in the Distribution table.
19:14:46 <Evolution> s/table/devroom/?
19:14:53 <smooge> devroom
19:15:24 <smooge> it is to go over what epel is, what it does, who uses it, what do people want from it and how to help it.
19:15:50 <nirik> excellent.
19:15:52 <smooge> i am trying to find the time exactly on Sunday of FOSDEM it is at
19:16:05 <bstinson> starts at 16:30
19:16:07 <orionp> thanks for that
19:16:14 <tibbs|w> Only about 3000 packages in EPEL5.
19:16:25 <bstinson> and goes until security kicks us out :)
19:16:33 <smooge> this is actually bstinson's brainchild I think. I am just a humble narrator
19:16:56 * nirik will not make it, but hopes everyone has a good discussion.
19:17:28 <smooge> i may not be at next 2 epel meetings due to travelling to and from fosdem
19:17:40 <nirik> alright.
19:17:47 <smooge> so if there are things that people would liek to be put on the meeting please let me know via email
19:17:52 <bstinson> smooge: thanks for organizing/moderating/wrangling
19:17:53 <smooge> thank you for your time
19:18:05 <nirik> #info FOSDEM epel round table - 16:30 in devroom. Be there!
19:18:21 <nirik> anything else on this?
19:19:07 <nirik> #topic Python3 in epel7
19:19:24 <nirik> It seems like we have stalled out a bit on this... wondering what we can do to move things forward again.
19:19:36 <nirik> python34 is in and macros are in I think...
19:19:49 <tibbs|w> Some of it was us trying to do some super magic macros, but that shouldn't be blocking anything.
19:20:04 <orionp> I've been getting a bunch of packages in...
19:20:27 <bstinson> do we just need reviews?
19:20:29 <nirik> oh, there are a number now. ;)
19:20:34 <orionp> https://bugzilla.redhat.com/showdependencytree.cgi?id=1294704&hide_resolved=1
19:21:19 <nirik> cool. I can try and fire up the review-o-tron this weekend and see if I can help.
19:21:29 <nirik> also, I am going to try and package python35 up
19:21:30 <Evolution> does this let people build against a python3 buildrequires currently?
19:21:43 <Evolution> or must we specify python34/python35 etc?
19:21:56 <orionp> BR python%{python3_pkgversion}-devel
19:22:20 <orionp> https://fedoraproject.org/wiki/PackagingDrafts:Python3EPEL
19:22:35 <orionp> minus the %py_package macro
19:22:55 <nirik> we should move that from drafts and make it clear what process people need to use
19:22:59 <orionp> that never got done, though tibbs and others have been working on a replacement
19:24:03 <tibbs|w> Yeah, that needs to move forward but it needs interested people to try and do some packages using the new macros to see how they need to change to accommodate reality.
19:24:05 <orionp> Also, we don't currently have a %python3_other_pkgversion - no one has yet packaged python35
19:24:21 <nirik> orionp: right. as above, I was going to try and do that...
19:24:25 <nirik> just haven't gotten to it yet
19:24:26 <tibbs|w> There's some fun stuff in there if you find RPM internals and lua fun....
19:24:42 <tibbs|w> If anyone wants to help with that, please let me know.
19:24:58 <nirik> cool.
19:25:08 <nirik> #info nirik to work on reviews and packaging python35.
19:25:23 <nirik> #info any folks interested in helping with python3 macros in epel7, talk to tibbs
19:25:29 <nirik> anything else on this?
19:26:05 <nirik> #topic Alternate install prefixes - orionp
19:26:12 <nirik> orionp: whats this about? :)
19:26:22 <orionp> https://lists.fedoraproject.org/archives/list/epel-devel%40lists.fedoraproject.org/thread/2DCRCGDJ3Q6VJTYPJNHFIBZNLIAJCZSX/
19:27:28 <orionp> Generally, how to make packaging up alternate versions easier
19:27:59 <nirik> yeah... this was perhaps orig what SCLs were for...
19:28:14 <nirik> in any case I don't think theres any easy road.
19:28:35 <orionp> It's definitely what SCLs are for, though I don't particularly like SCLs
19:28:40 <tibbs|w> I wish we could just allow multiple versions of a package to be installed as long as they don't conflict.
19:28:49 <tibbs|w> And then just make sure they don't conflict.
19:29:10 <tibbs|w> RPM doesn't care, but dnf/yum treats the kernel magically.
19:29:15 <orionp> Well, the jist of my question is how to best make them not conflict
19:29:25 <tibbs|w> And for EPEL, well, there's not a whole lot you can do about how yum works.
19:29:54 <nirik> orionp: I fear the best answer is the invasive patch at this point at least.
19:30:11 <orionp> for some things it's easy to add a suffix, not so for cmake
19:30:53 <tibbs|w> So cmake insists on calling itself as "cmake"?
19:30:55 <nirik> yeah, some packages are not at all easy for that.
19:31:10 <tibbs|w> Can that be fixed with a wrapper that sets $PATH?
19:31:49 <orionp> sure, but the guidelines don't really allow installing elsehwere do they?
19:32:02 <tibbs|w> Well, guidelines can be changed.
19:32:29 <tibbs|w> I wonder what the FHS has to say about this.
19:32:39 <tibbs|w> Plus, Fedora does allow libexec.
19:32:53 <tibbs|w> Not that I think it's a particularly good solution here.
19:33:10 <tibbs|w> I think FHS forbids subdirectories in /usr/bin, which would be one solution.
19:33:32 <tibbs|w> And it's not at all uncommon for executables to be in %libdir or /usr/share and to have symlinks.
19:33:40 <tibbs|w> Well, under those directories.
19:34:04 <nirik> perhaps something under libdir would work.
19:34:38 <smooge> orionp, something like /opt/epel/<tree>/?
19:34:58 <orionp> smooge: that's kind of what I was hoping for
19:35:39 <nirik> I suppose it's possible, would need to register epel there...
19:35:43 <smooge> there would still need to be a way to have multiple packages in that tree and how to 'link' them back
19:35:47 <nirik> or use /opt/fedora/epel/ or something
19:36:20 <orionp> more cmake it would just be /usr/bin/cmake34 -> /opt/epel/cmake34/bin/cmake
19:36:39 <Evolution> that way feels like madness.
19:36:41 <orionp> s/more/for/
19:36:58 <orionp> It's basically SCLs
19:37:12 <orionp> but with different "activation"
19:37:15 <nirik> well, no, it doesn't have any of the scl wrapper around it right?
19:37:32 <Evolution> you could almost abuse alternatives for that.
19:37:41 <Evolution> use alternatives to set/create the link, ala java.
19:37:59 <tibbs|w> Thing is, if it's basically SCLs, aren't we back in the same place?
19:38:01 <nirik> alternatives would work if you had cooperation from the rhel cmake maintainer to make changes, not if not
19:38:02 <orionp> It's not the activation as much as where to install it
19:38:32 <tibbs|w> Then again, if the differences are positive, then maybe we've finally found someone willing to try and make something more acceptable than SCLs.
19:39:50 * nirik doesn't think we are going to solve this here and now. ;) Perhaps continue to discuss on list / try things out?
19:40:01 <tibbs|w> Quite true.
19:40:06 <smooge> yes agreed.
19:41:07 <nirik> #topic More arch support - Evolution
19:41:15 <nirik> Evolution: what did you want to bring up here?
19:41:23 * nirik goes to grab more coffee, back in a min
19:42:53 <tibbs|w> Well, over the span of this meeting I've rebuilt EPEL5.
19:43:02 <tibbs|w> Well, all but 10 huge packages.
19:43:11 <nirik> ha. cool..
19:44:13 <nirik> anyhow, on more arches we have i686, aarch64 and s390. I think all those are doable, i686 likely would be the easiest as we have builders already that can do it.
19:46:13 <nirik> we just need to get cycles from pbrobinson I think. :)
19:46:33 <nirik> #topic New package notification - orionp
19:46:51 <nirik> orionp: this was just like the weekly job we have mailing devel list about new/orphaned/etc packages ?
19:47:13 <orionp> no - thenewhotness for EPEL only packages
19:47:35 <nirik> ahhhh....
19:48:13 <orionp> I think I asked about it before...
19:48:19 <nirik> I think we will need to talk with threebean / pingou and see whats involved.
19:48:27 <nirik> might be we treat epel as another distro.
19:49:01 <orionp> Right, I managed to create "Fedora EPEL" in anitya
19:49:20 <tibbs|w> Are there really that many packages where this would be needed?
19:49:21 <orionp> but that needs to get turned into bugs/notifications I think
19:49:22 <nirik> then we need to have new-hotness deal with it I guess.
19:49:42 <orionp> Well, all the python3- packages....
19:49:50 <orionp> that are in RHEL7
19:49:56 <smooge> tibbs|w, congrats what were the 10 leftovers?
19:50:03 <tibbs|w> Down to five, actually.
19:50:12 <orionp> so no, not lots, but still would be helpful
19:50:21 * nirik nods.
19:50:24 <smooge> ok Evolution is afk so I will try to pick this up
19:50:29 <nirik> I see you have a ticket filed already.
19:50:35 <tibbs|w> root, ligbuestfs, nordugrid, perl-Gearman-Client-Async, thunderbird-lightning.
19:50:42 <tibbs|w> And root just finished.
19:51:07 <orionp> tb-lightning - hoo boy
19:51:25 <tibbs|w> I suspect some of them are simply hung.
19:51:53 <nirik> tb-lightning might no longer be needed. it was pulled into thunderbird at some point
19:52:28 <nirik> #topic Open Floor
19:52:33 <tibbs|w> Anyway, 178 failures, plus probably perl-Gearman-whatever because I think its test suite dies.
19:52:37 <tibbs|w> Well, just hangs.
19:52:43 <nirik> anything we didn't cover / misc stuff?
19:52:46 <smooge> nirik, I can cover alternative architectures
19:52:49 <nirik> I had a few info items
19:52:59 <smooge> but will do so after your info itmes
19:53:00 <tibbs|w> This does bring up the question of whether anything ever actually gets pruned from EPEL.
19:53:16 <nirik> smooge: oh, you had more info on it, sure...
19:53:20 <tibbs|w> Because from what I've seen, there is a whole lot that is basically unmaintained.
19:53:24 <Evolution> sorry. back. phone call
19:53:37 <smooge> ok will turn over to Evolution
19:53:41 <nirik> #topic More arch support - Evolution
19:53:59 <nirik> tibbs|w: they are, but there's not a very concerted effort to prune often. ;(
19:54:01 <Evolution> ppc64(le) appears to be covered.
19:54:22 <nirik> yep it's all in and working
19:54:38 <Evolution> I'm not sure if the plan for aarch64 is to wait for the official aarch64 build from rhel
19:54:50 <smooge> correct
19:55:06 <Evolution> for arm32, that's not an issue, but we have more to clean up on our side before that's a consideration.
19:55:18 <nirik> plan? ha.
19:55:45 <nirik> IMHO we should look at i686 next... but I would defer to pbrobinson as he's been doing the work
19:56:04 <Evolution> he's expressed some concern on i686
19:56:19 <Evolution> mostly around timing/conflicts for x86_64 I believe
19:56:50 <Evolution> the current plan is to discuss that with him at fosdem, possibly during/after the roundtable smooge is working.
19:56:54 <nirik> ok
19:57:05 <Evolution> once we have an idea for what he sees as problem sets, we can work on solutions
19:57:24 <smooge> We are to take him to a couple of bars afterwords, get his signature on some documents and then implement.
19:57:52 <nirik> excellent.
19:58:09 <nirik> anything else on this?
19:58:21 <Evolution> I don't believe so.
19:58:25 <smooge> I am working on getting CentOS mirrored into Fedora infrastructure correctly
19:58:42 <Evolution> smooge: does/will that include the altarch bits?
19:58:59 <smooge> I am hoping so
19:59:01 <dgilmore> sorry I am here now
19:59:13 <nirik> #topic Open Floor
19:59:17 <nirik> ok, back to open floor. ;)
19:59:17 <smooge> since that would make rebuilding i686 so much easier :)
19:59:25 <orionp> Did we ever make progress on a repoclosure report for EPEL?
19:59:38 <nirik> #info ansible 2.0.0.x is in testing, will stay there for a while until things stablize and go into updates.
19:59:49 <nirik> orionp: nope, I haven't at least
19:59:51 <dgilmore> smooge: we can sync i686 centos, the tricky bit is entirely when rhel point releases come out
20:00:10 <nirik> #info latest python-sqlalchemy is in epel7-testing. Please test for any regressions. :)
20:01:12 * orionp is going to need to pack a lunch for these meetings....
20:01:14 <smooge> nirik, is that tied to ansible also or other packages?
20:01:42 <nirik> the python-sqlalchemy? no, it's completely seperate.
20:02:21 <nirik> anything else? we are already over... ;)
20:02:25 <nirik> if not will close in a min
20:02:39 <tibbs|w> Ugh, thunderbird-lightning builds one file at a time.
20:02:56 <tibbs|w> I'll post the list of failures once this finishes.
20:03:11 <nirik> cool. Thanks.
20:03:22 <nirik> thanks for coming everyone. ;) Great to see some interested folks...
20:03:26 <nirik> #endmeeting