17:59:39 <smooge> #startmeeting EPEL (2018-03-14)
17:59:39 <zodbot> Meeting started Wed Mar 14 17:59:39 2018 UTC.  The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:59:39 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:59:39 <zodbot> The meeting name has been set to 'epel_(2018-03-14)'
17:59:39 <smooge> #meetingname EPEL
17:59:39 <zodbot> The meeting name has been set to 'epel'
17:59:39 <smooge> #topic aloha
17:59:39 <smooge> #chair avij bstinson Evolution nirik smooge
17:59:39 <zodbot> Current chairs: Evolution avij bstinson nirik smooge
17:59:54 <smooge> Good morning.. I started a little early
18:00:00 <smooge> happy pi day
18:00:21 <bstinson> \o/ epel meeting
18:00:34 <nirik> morning
18:01:18 <smooge> ok this will be a short meeting as I didn't have an agenda
18:01:23 <smooge> #topic Agenda
18:01:39 <smooge> #info The agenda is a lie.
18:01:53 <smooge> does anyone have something to discuss this week?
18:02:01 * carlwgeorge waves
18:02:20 <smooge> #info epel.io was transferred to Fedora. We need to use it
18:02:26 <smooge> thanks carlwgeorge
18:02:28 <nirik> thanks for the blog post on devtoolset smooge
18:02:30 <tmz> Is devtoolset in koji on the agenda already?
18:02:39 <smooge> tmz, it was last week
18:02:40 <carlwgeorge> if there's time i'd like to discuss python34/36
18:02:50 <smooge> tmz what do you need
18:03:11 <smooge> #info Agenda item: devtoolset in koji
18:03:20 <smooge> #info Agenda item: Python34/36
18:03:27 <tmz> smooge: I am curious about testing it with mock and koji to ensure things work as similarly as possible between the two.
18:03:29 <smooge> #info Agenda item: EPEL advertising
18:03:47 <smooge> #topic devtoolset in kojo
18:03:54 <nirik> I don;'t see why they would be any different.
18:03:54 <smooge> cujo
18:03:56 <tmz> The mock configs have an includepkgs statement to limit to devtoolset.
18:04:25 <smooge> ah. nirik did you set it up as the same?
18:04:26 <tmz> I don't know what the koji configs look like.  We also only have mock configs for x86 and x86_64.
18:05:08 <smooge> x86?
18:05:11 <nirik> I haven't limited anything yet... not sure how to or if we need to.
18:05:12 <tmz> I never saw any solid information on whether there were stable setups for other arches.  The last I saw they were in testing, so I didn't add them to the mock configs upstream
18:05:16 <tmz> i386
18:05:39 <nirik> x86_64, ppc64le, ppc64, aarch64 are all valid
18:05:46 <smooge> I would not use ppc64
18:05:51 <smooge> it is just gdb
18:06:01 <nirik> ah, ok
18:06:05 <tmz> Does anyone know where the centos repos are for those other arches?
18:06:14 <smooge> I don't think there are any
18:06:24 * nirik doesn't know
18:06:27 <smooge> but I will bow to bstinson on that
18:06:38 <smooge> or avij
18:06:56 <tmz> Okay.  I suppose the best that can be done then is include something in the eventual wiki docs about that?
18:07:08 <smooge> OK will do so.
18:07:22 <bstinson> alternate arches live at http://mirror.centos.org/altarch/
18:07:38 <smooge> are there scl sig builds in that?
18:07:59 <bstinson> as of a couple weeks ago...YES!
18:08:04 <smooge> cool
18:08:06 <tmz> smooge: I think you noted that this was also only in koji for epel7, not el6?
18:08:19 * nirik nods
18:08:23 <bstinson> we're working through the centos-release packages (to install the appropriate repo files) at the moment
18:08:37 <smooge> Correct. There were requests long ago about EL6 but I wanted to see if they were still wanted
18:09:00 <smooge> bstinson, could you work with tmz on any of the mock config fixes for it?
18:09:06 <bstinson> definitely
18:09:10 <smooge> thanks
18:09:25 <tmz> I don't have any pressing need for it, but the only time I wanted to use newer gcc was on el6.  I don't know if there are current epel packagers in that boat or not though.
18:09:36 <tmz> Thanks smooge and bstinson.
18:09:52 <smooge> tmz, do you still need it?
18:10:30 <smooge> if you do, it should take me about a week to get it mirrored in x86_64 on EL6.. I don't know about the others
18:10:37 <tmz> smooge: Not for anything I maintain for el6 in epel.  I might use it for work/personal builds, but there I'd use mock anyway.
18:10:37 * nirik read recently that rhel6 is going end of life in only 2.5 more years. :)
18:10:53 <smooge> yeah but it is still the most used EPEL
18:11:06 <tibbs> Well 5 was until recently.
18:11:06 <smooge> which is todays blog
18:11:09 <tmz> nirik: They grow up so quickly.
18:11:19 <smooge> no 5 was not for a long time
18:11:36 * nirik looks at his rawhide laptop... hard to imagine the rhel6 days. :)
18:11:37 <tibbs> Not according to my mirror stats, at least.
18:12:04 <smooge> ah ok
18:12:37 <tibbs> Now that it's in archive it's tough to compare, since there are fewer mirrors of archive so we see more traffic.
18:13:00 <smooge> I expect every mirror has its own usage patterns
18:13:18 <tibbs> Still about a request per second for EPEL5 repomd.xml on all architectures.
18:13:20 <smooge> I only see the 'hey which mirror should I go to' not the 'I NEED OMNOMNOM'
18:13:24 <tibbs> But anyway, sorry for derailing.
18:14:02 <smooge> no problem.. we need a good rail
18:14:11 <tmz> Ha
18:14:14 <smooge> ok anything more on DTS at the moment?
18:14:19 <tibbs> That could be misinterpreted....
18:15:09 <smooge> if so I don't have any idea how but I have a tendency to not see the woods for the trees in comments
18:15:12 <tmz> I'm good there.  I'll talk to bstinson and see about getting epel-7-{aarch64,ppc64le}.cfg configs updated.
18:15:19 <smooge> ok cool.
18:15:38 <smooge> #info IF YOU WANT DTS IN EL6 FOR A PACKAGE LET US KNOW
18:15:50 <smooge> #topic Python34/36
18:15:58 <smooge> carlwgeorge, you are up
18:16:21 <carlwgeorge> I was curious what the timeline was for switching %python3_pkgversion from 34 to 36
18:16:39 <carlwgeorge> Also, what's the plan for handling the /usr/bin/python3 symlink
18:17:02 <tibbs> Who is actually running the epel python3 effort these days?
18:17:12 <tibbs> I recall that it was Orion before, but he hasn't had much time lately.
18:17:28 <nirik> cstratak perhaps?
18:18:10 <nirik> and yeah, perhaps we should do that when the next minor comes out?
18:18:29 <tibbs> Or should you ever do it?
18:18:48 <nirik> we planned to...
18:19:21 <carlwgeorge> i can't imagine people using python34 will be happy when their python3 command disappears
18:19:34 <tibbs> It's one of those things that once it gets into EPEL, you can't change it without breaking someone.
18:19:51 <nirik> https://fedoraproject.org/wiki/User:Bkabrda/EPEL7_Python3
18:19:55 <nirik> look at lifecycle
18:20:02 <tibbs> Sometimes I wonder if the symlink itself shouldn't be in a separate package.
18:20:18 <nirik> at a certain point at time, annoucement is made that python34 is to be retired and python35 is to be *the* one - at this point, python34 and python35 are rebuilt
18:20:20 <tibbs> And then let that use alternatives or something horrible.
18:20:51 <smooge> so 1 we need to find who is currently running it
18:21:10 <smooge> 2. if no one is, get someone *cough* carlwgeorge didn't step back
18:21:11 <nirik> well, if they want python34, then they should call that I would think?
18:21:15 * cstratak waves
18:21:23 <smooge> or cstratak
18:21:34 <carlwgeorge> nirik: I would agree, but not everyone sees it like that
18:21:48 <smooge> 3. we are at a 7.5 beta -> 7.5 so I think we do it then
18:22:15 <cstratak> I haven't followed so far, so what is the issue with EPEL exactly?
18:22:35 <nirik> cstratak: when to switch from python34 to python36
18:22:41 <tibbs> Maybe using the RHEL minor version transition points as a flag day for potentially breaking EPEL changes isn't a bad idea.
18:22:54 <carlwgeorge> For comparison, IUS just stopped including python3 and just left the python3.x command, so the packages would be fully parallel installable, but we're getting negative feedback about that now. https://github.com/iuscommunity/packaging/issues/1
18:22:55 <cstratak> honestly I didn't have too much time to plan a roadmap
18:23:17 <nirik> tibbs: the downside is the centos not being updated at the same time...
18:23:18 <cstratak> I'd like to resolve the guidelines issue first
18:23:29 <nirik> and also that we often don;;t know when exactly it will be
18:23:40 <nirik> cstratak: guidelines issue?
18:23:50 <tibbs> Well yeah, Red Hat keeping secrets will always cause problems.
18:23:50 <cstratak> there is an open ticket regarding that
18:24:06 <cstratak> on FPC, let me find it
18:25:14 <cstratak> https://pagure.io/packaging-committee/issue/567
18:25:15 <smooge> I would go with something like python34-fing-symlink and python36-fing-symlink
18:25:44 <carlwgeorge> tibbs, you just hit on two of the possibilities that IUS is considering, alternatives and conflicting symlink subpackages
18:27:24 <smooge> cstratak, ah ok. I think that they are wanting that to be an EPSCO issue and we didn't realize it was there
18:27:50 <nirik> yeah, I had no idea. :)
18:28:03 <tibbs> I've sort of lost track of that whole thing.  And FPC is going through a pretty big change at the moment (replacing a number of members) so we've been behind.
18:28:35 <tibbs> But in general FPC has enough going on just keeping up with Fedora's python.
18:28:45 <smooge> cstratak, ok lets see if we can work that out . I will add that to my worklist to get back to you next week.
18:28:53 <tibbs> The EPEL stuff needs to be decided and written into the EPEL guidelines.
18:28:59 <nirik> what is the status there...
18:29:03 <nirik> it's kinda hard to tell.
18:29:15 <tibbs> And if the main python guidelines need mention of that then of course let FPC know.
18:29:58 <cstratak> smooge, thanks.
18:30:04 <tibbs> And I think the idea was that since FPC asked that EPSCo handle it, someone who wanted it to happen would bring the issue before EPSCo.  I guess that never happened.
18:30:33 <smooge> what guidelines are you looking for?
18:30:43 <nirik> so basically we need to have someone polish up the proposed rules, bring it to us here and approve it?
18:30:59 <nirik> (I'm asuming there might be some changes due to recent macros and other stuff over the last few years)
18:31:24 <tibbs> I think what this all needs is someone to figure out just how on earth a packager is supposed to support however many python stacks EPEL is going to have.
18:32:15 <carlwgeorge> i believe the intention was utilizing %python3_other_pkgversion alongside %python3_pkgversion
18:32:21 <orc_fedo> nirik the F25 'recommends', and the --define "rpmmacrodir %{_sysctldir}/../rpm/macros.d/"   \ is about al the heartburn I've had to face
18:32:43 <cstratak> carlwgeorge, that is correct
18:32:45 <nirik> right, so py2 and py3(now) and py3(other) are all...
18:33:18 <nirik> orc_fedo: in the python3 guidelines, or in general you mean?
18:33:20 <tibbs> That assumes there will only be three stacks (py2, py3-prime and py3-secondary).
18:33:33 <orc_fedo> nirik as to macro changes in general
18:33:35 <smooge> tibbs, so we were planning to support only 2 stacks
18:33:54 <smooge> py2 and py3x
18:34:26 <smooge> but that would require us to figure out a way to archive off old py3(x-1)
18:34:46 <smooge> because of the way we build stuff it would go away
18:35:16 <smooge> but that was N years ago.. and I am not sure it works
18:35:18 <cstratak> ideally one for python2 and two for python3 during transitioning (which is the situation now), and at some time the older python3 stack should phase out.
18:35:56 <smooge> cstratak, yes that is what I think would be workable. When we put in the new one, we don't build for the old one anymore.. but keep it there until it is phased out
18:36:54 * nirik nods
18:38:18 <smooge> so work on a guideline like that and publish to the list later this week
18:38:26 <smooge> I will put that on my list
18:38:39 <smooge> cstratak, tibbs would you be available to give feedback when you can?
18:38:54 <tibbs> Of course.
18:38:55 <nirik> should be possible to test things since 34 and 36 are both in
18:39:15 <tibbs> Just CC me on whatever and bug me if I don't respond quickly.
18:39:26 <cstratak> smooge, sure
18:39:44 <smooge> ok thanks.. I will put that as something for my next item
18:39:52 <cstratak> smooge, irc, emails, cc on bugs and issues work for me.
18:40:08 <smooge> ok will do so
18:40:17 <smooge> anything else on this? does htis help carlwgeorge ?
18:40:57 <carlwgeorge> cstratak: in the meantime, would you be open to pull requests to bump python34/36 up to the latest?  3.4.5->3.4.8 and 3.6.3->3.6.4?
18:42:29 <cstratak> carlwgeorge, yep of course
18:42:44 <nirik> that would be lovley...
18:43:01 <nirik> when is 3.7 out?
18:43:20 <cstratak> carlwgeorge, I was planning to do that but well, haven't found the time yet. A PR would be great.
18:43:27 <cstratak> 3.7 is around june if I recall correctly
18:43:35 <carlwgeorge> i'll probably do that soon.  i maintain those in ius, and i honestly wouldn't mind being able to eventually have epel's obsolete ius's and just help maintain it there.
18:43:36 <nirik> https://www.python.org/dev/peps/pep-0537/
18:43:42 <nirik> yeah, 6-15ish
18:43:57 <cstratak> anyway gotta run. Thanks a bunch for the initiative.
18:44:02 <smooge> ok thanks
18:44:05 <smooge> next up
18:44:13 <smooge> #topic EPEL Advertising
18:44:24 <smooge> So I am trying to write a blog a day this week on EPEL.
18:44:52 <smooge> Do people have topic ideas for me to cover?
18:45:19 <smooge> How to test and give karma to a package is one I am working on today
18:45:50 <smooge> but I would like others for the next week or so
18:45:58 <avij> yes, letting people know that there's epel-testing would be nice
18:46:09 <tibbs> Explaining the stability promises and how to find out if changes are coming in a package upon which you depend?
18:46:38 <smooge> OK I can do the first.. the second one I am never sure on :)
18:46:39 <nirik> How to file bugs? (some people don't know where)
18:47:04 <tibbs> I think BugURL: is set even for EPEL packages now.
18:47:11 <nirik> smooge: oh, BTW, I still have a tab open on your new how to get a package added to epel thing... someday I will review it. I promise. Someday.
18:47:27 <nirik> tibbs: yep. if they knew to look for it there.
18:47:36 <smooge> my browser has crashed too many times to have any tabs open
18:47:37 <carlwgeorge> smooge: how to 1) request an epel build of a fedora package and 2) how to request an epel7 build of an epel6 package
18:47:44 <avij> I would like to see a "please use epel-testing on your development servers, and when necessary, give good/bad karma to packages to make sure only good packages end up to stable, and that they end up there without unnecessary delays" mention in there
18:47:52 <carlwgeorge> i get asked that routinely
18:48:13 <smooge> nirik, can you give me a link to that tab :)?
18:48:22 <nirik> uh, looking...
18:48:22 <smooge> I can give it to carlwgeorge
18:48:29 <orc_fedo> smooge: there are some ancient and untestable items in the epel6 and epel 7 testing report -- binaries and SRPMs and logs long gone
18:48:39 <nirik> https://fedoraproject.org/wiki/Getting_a_Fedora_package_in_EPEL
18:48:54 <carlwgeorge> nice, i'll share this
18:49:01 <nirik> but looking at it now, could we advise changing that to filing a bug instead of email?
18:49:04 <tibbs> That page is from an earlier time.
18:49:23 <smooge> carlwgeorge, well review it before you do so
18:49:42 <tibbs> These days we've kind of blurred any distinction between different maintainers for different branches.
18:49:49 <nirik> yeah.
18:49:57 <carlwgeorge> yeah that jumped out at me, no more per-branch maintainers
18:50:10 <nirik> and if we have people file a bug, we could make a template automatically from a link...
18:50:11 <tibbs> Which is unfortunate because there are package which I thought I never wanted to maintain in EPEL, but now I can't tell if I'm supposed to maintain there.
18:50:17 <smooge> tibbs, I can fix it if I know how it is done these days
18:50:47 <tibbs> smooge: I don't think you can.  The distinction simply does not exist.
18:50:58 <nirik> you can still have epel bugs be assigned to someone else...
18:51:09 <smooge> .whoowns clamav
18:51:09 <zodbot> smooge: sergiomb (robert in Fedora EPEL)
18:51:21 <smooge> so that isn't right anymore ^^^
18:51:28 <nirik> I bet thats still trying to hit pkgdb
18:52:02 <carlwgeorge> oh i did a thing for that, i'm the bugzilla default assignee for git-lfs, it's not using pkgdb anymore
18:52:04 <smooge> <head desk> I have been telling people for the last N months
18:52:33 <carlwgeorge> i had to do a pull request to some pagure repo, looking...
18:53:00 <nirik> right. there's a override file...
18:53:03 <carlwgeorge> https://pagure.io/releng/fedora-scm-requests/blob/master/f/rpms/git-lfs
18:53:12 <nirik> https://pagure.io/releng/fedora-scm-requests
18:53:45 <smooge> ... wonders how we are going to deal with thousands of module branches with different assignees
18:54:08 <smooge> decides that he needs to focus on the meeting
18:54:27 <smooge> ok so I will work on the above ideas into posts
18:54:27 <smooge> thanks
18:54:33 <smooge> #topic Open Floor
18:54:50 * nirik will file a bug to get zodbot updated.
18:54:53 <smooge> OK I have filled up this hour pretty well. I see one open floor item
18:55:05 <smooge> <orc_fedo> smooge: there are some ancient and untestable items in the epel6 and epel 7 testing report -- binaries and SRPMs and logs long gone
18:55:22 <tibbs> More python2 compat packages are in; the final pushes to stable should happen tomorrow if bodhi cooperates.
18:55:24 <orc_fedo> smooge: ayuo
18:55:29 <orc_fedo> ayup
18:55:35 <orc_fedo> I filed a bug, etc
18:55:48 <smooge> can you give me a bug number.. I will see what needs to be done on it
18:55:59 <smooge> tibbs, thanks
18:56:11 <tibbs> As always, if someone wants more of those python2 stub packages, add them to https://fedoraproject.org/wiki/User:Tibbs/EPELPythonStubPackages and I'll take care of them.
18:56:44 <orc_fedo> smooge: 4 email in your box already but hokay
18:57:02 <smooge> ok thanks
18:57:04 <smooge> will look
18:57:15 <smooge> I am going to call this meeting done.
18:57:19 <tibbs> I got one weird confused bug about one of the stub packages, telling me that I needed to update to a newer version.  I still don't know how the person became confused by that.
18:57:42 <smooge> they never responded did they?
18:58:06 <smooge> I looked and have no idea where they got that package from
18:58:09 <orc_fedo> smooge> <head desk> I have been telling people for the last N months
18:58:11 <orc_fedo> nirik> I bet thats still trying to hit pkgdb
18:58:24 <tibbs> I think they just expected that package to track what's in Fedora somehow.
18:58:25 <orc_fedo> why not null route this and forece bugs to appear?
18:58:58 <smooge> I don't know. That would be something for the infrastructure meeting tomorrow
18:59:44 <smooge> so I will see about putting it on the agenda since I do that also :)
18:59:47 <nirik> null route what? pkgdb? well, zodbot has a cache when it starts... and we are still porting some things from it so we don''t want to turn it off right now... but we should set a date
19:00:18 <smooge> ok I am closing this meeting out
19:00:21 <smooge> #endmeeting