16:06:54 <dgilmore> #startmeeting FESCO (2017-01-13)
16:06:54 <zodbot> Meeting started Fri Jan 13 16:06:54 2017 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:06:54 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:06:54 <zodbot> The meeting name has been set to 'fesco_(2017-01-13)'
16:06:54 <dgilmore> #meetingname fesco
16:06:55 <zodbot> The meeting name has been set to 'fesco'
16:06:55 <dgilmore> #chair maxamillion dgilmore jwb nirik paragan jsmith kalev sgallagh Rathann
16:06:55 <zodbot> Current chairs: Rathann dgilmore jsmith jwb kalev maxamillion nirik paragan sgallagh
16:06:57 <dgilmore> #topic init process
16:07:03 <kalev> .hello kalev
16:07:04 <dgilmore> hi all who is here
16:07:04 <zodbot> kalev: kalev 'Kalev Lember' <klember@redhat.com>
16:07:10 <nirik> .hello kevin
16:07:11 <zodbot> nirik: kevin 'Kevin Fenzi' <kevin@scrye.com>
16:07:17 <sgallagh> .hello sgallagh
16:07:17 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
16:07:22 <dgilmore> #info appologies for failing to get the agenda out :(
16:07:25 <jsmith> .hello jsmith
16:07:25 <zodbot> jsmith: jsmith 'Jared Smith' <jsmith.fedora@gmail.com>
16:07:27 <sgallagh> (My attention is split today, so please forgive me if my responses are slow)
16:07:36 <dgilmore> sgallagh: it should be quick
16:07:53 <paragan> .hello pnemade
16:07:54 <zodbot> paragan: pnemade 'Parag Nemade' <pnemade@redhat.com>
16:08:02 <dgilmore> okay lets get moving
16:08:36 <dgilmore> #topic #1667 F26 System Wide Change: Switch OpenLDAP from NSS to OpenSSL
16:08:44 <dgilmore> https://pagure.io/fesco/issue/1667
16:09:10 <dgilmore> .fesco 1667
16:09:11 <zodbot> dgilmore: Issue #1667: F26 System Wide Change: Switch OpenLDAP from NSS to OpenSSL - fesco - Pagure - https://fedorahosted.org/fesco/ticket/1667
16:09:21 <sgallagh> I'm surprised it took this long. The OpenLDAP maintainer has wanted to do this for a couple years now.
16:09:27 <dgilmore> fedmsg needs updating for the URL
16:09:51 <kalev> +1
16:10:00 <nirik> I can see why they would... but I won't stand in the way. +1
16:10:03 <sgallagh> +1, looks like they have already looped in all the relevant maintainers.
16:10:12 <dgilmore> +1
16:10:15 <paragan> +1
16:10:18 <sgallagh> nirik: Sorry, could you elaborate?
16:10:51 <nirik> I understand why they would want to do this? I approve?
16:11:00 <nirik> not sure what else I can elaborate on...
16:11:40 <sgallagh> It was the "but I won't stand in the way" part that was confusing
16:11:41 <jsmith> +1
16:11:51 <nirik> sorry, that was poor phrasing
16:12:19 <sgallagh> No worries. Just didn't want to see any legitimate concerns go unheard
16:13:17 <dgilmore> proposed #accepted F26 System Wide Change: Switch OpenLDAP from NSS to OpenSSL is accepted (6:+ 0:0 0:-)
16:13:50 <dgilmore> #accepted F26 System Wide Change: Switch OpenLDAP from NSS to OpenSSL is accepted (6:+ 0:0 0:-)
16:14:20 <dgilmore> #topic #1666 F26 System Wide Change: pkgconf as system pkg-config implementation
16:14:26 <dgilmore> https://pagure.io/fesco/issue/1666
16:14:31 <dgilmore> .fesco 1666
16:14:32 <zodbot> dgilmore: Issue #1666: F26 System Wide Change: pkgconf as system pkg-config implementation - fesco - Pagure - https://pagure.io/fesco/issue/1666
16:15:31 <sgallagh> I did a brief investigation of this for the Base Runtime team. In my opinion, the risks of making this default from a technical standpoint are minimal. So it's a purely political/ideological decision.
16:15:54 <kalev> I'm a bit on the fence with this one
16:16:16 <kalev> I don't think we've had any major issues with freedesktop pkg-config so far which would force us to switch
16:16:36 <kalev> and also not sure what benefits the "pkgconf" implementation would give us compared to the freedesktop implementation
16:17:03 <nirik> well, they have a library interface (not sure how much anyone would use it tho)
16:17:03 <sgallagh> kalev: There's a library that one can use instead of screen-scraping the CLI tool, among other things.
16:17:08 <sgallagh> I can see that being useful.
16:17:14 <nirik> and they are active, where the other upstream doesn't seem to be at all.
16:17:16 <kalev> I guess it depends on which one of them becomes more prevalent, but it's unclear at this point which is going to win
16:17:18 <dgilmore> I would like to hear jakub's thoughts on it
16:17:45 <sgallagh> kalev: I kind of get the impression that the freedesktop pkgconfig is basically prevalent by inertia rather than specific value
16:17:52 <kalev> sgallagh: sure, but putting the library in fedora is orthagonal to switching /usr/bin/pkg-config to a new implementation
16:18:07 <kalev> as in, the library can still go in fedora and we can reap all the benefits of having it in
16:18:29 <sgallagh> kalev: Well, regardless of how unlikely it looks, I really wouldn't want the two tools to be providing differing output.
16:18:50 <sgallagh> If pkgconf (under active development) starts providing more content, then they're going to differ.
16:19:04 <sgallagh> Even if they are still arguably "compatible"
16:19:37 <sgallagh> Put another way: if we have both in the distro, we should limit the requirement that the .pc format only offers capabilities supported by both.
16:19:45 <sgallagh> At which point... why bother including pkgconf?
16:21:25 <nirik> yeah, I guess I am +1 to the change...
16:21:31 <dgilmore> seems like we should postpone the decision here and ask questions on list
16:21:35 <sgallagh> So, I'm +1 to obsoleting the existing one.
16:21:42 <dgilmore> I do not have a stong opinion
16:22:08 <nirik> we do need to decide before the mass rebuild...
16:22:19 <dgilmore> we do
16:22:23 <kalev> I'd mostly like to do what other distributions decide to do, so that we wouldn't needlessly differ
16:22:25 <sgallagh> Upstream claims full compatibility with properly-written .pc files, but does not claim compatibility with some of pkg-config's tolerant behavior of badly-formatted files.
16:22:39 <dgilmore> it is in 3 weeks
16:22:39 <sgallagh> Which seems like if we discover such issues, we'd be doing someone a favor anyway
16:23:00 <sgallagh> kalev: Remind me what the Four Foundations are again? ;-)
16:23:01 <dgilmore> sgallagh: and needing more time to clean up the fallout
16:24:11 <kalev> anyway, I am not neccesarily against it, just don't see much of a point of switch right now
16:24:33 <dgilmore> kalev: same
16:24:34 <jsmith> dgilmore: Fun, Functional, Fantastic, and Food :-)
16:24:45 <sgallagh> kalev: Not that it's a tremendous advantage, but dropping the glib2 dependency might help minimization and modularization efforts as well
16:24:54 <dgilmore> jsmith: you forgot Fitness
16:25:19 <sgallagh> jsmith: Also Facetious
16:26:15 <dgilmore> so we do not have enough positive votes for this to pass
16:26:17 <sgallagh> If we defer, what are the exact questions you want answered?
16:27:12 <jsmith> dgilmore: I'm not sure I've seen a proposal yet to actually vote on...
16:27:23 <dgilmore> jsmith: well the change itself
16:27:32 <jsmith> dgilmore: Are we voting on "let the package in, but not make it the defailt" or "let the package in and make it the default"?
16:27:48 <nirik> the change replaces the existing one with the new one.
16:27:56 <dgilmore> jsmith: well the change is to make it the default
16:27:58 <sgallagh> Well, "make it the default" I think has to be "Obsoletes:"
16:28:15 <sgallagh> The package is already in u-t, albeit without the compat symlink
16:28:21 <dgilmore> the package is already in or soon will be
16:28:30 <jsmith> Exactly... I need more details before I feel comfortable voting
16:28:31 <dgilmore> but you have to take extra steps to use it
16:28:45 <dgilmore> jsmith: so thats a question to ask on the list :D
16:29:02 <nirik> the plan was to enable it for mass rebuild and retire the old one after if no big issues are found.
16:29:47 <dgilmore> outside of its being actively developed, I have not seen anythings selling why its better
16:29:55 <jsmith> OK, I guess I'm +1, as long as I don't have to clean up the mess if the mass rebuild goes horribly wrong...
16:29:55 <dgilmore> dropping glib2 could be a + also
16:30:25 <dgilmore> maybe active development is enough?
16:30:47 <sgallagh> Active development would be enough to me, honestly.
16:31:04 <dgilmore> sgallagh: so you are +1
16:31:10 <dgilmore> jsmith: is +1
16:31:15 <sgallagh> I feel like that's exactly why we made the python-pillow change a while back.
16:31:24 <dgilmore> I am kinda a 0.25
16:31:31 <nirik> http://pkgconf.org/features.html
16:32:22 <dgilmore> will it still need rebuild for gcc version changes?
16:32:35 <dgilmore> not that its a huge deal
16:32:52 <dgilmore> but it sometimes is forgotten until we hit brokendeps
16:33:03 <kalev> dgilmore: it's libtool that needs a rebuild for gcc version changes, not pkg-config
16:33:15 <dgilmore> kalev: oh right
16:33:22 <dgilmore> I always mess the two up
16:33:54 <dgilmore> .whoowns pkg-config
16:33:54 <zodbot> dgilmore: No such package exists.
16:34:10 <dgilmore> so we are at 20 minutes here
16:34:12 <praiskup> I think it is named pkgconfig
16:34:16 <nirik> so I guess without enough votes we ask more questions on list and check back next week?
16:34:18 <dgilmore> .whoowns pkgconfig
16:34:18 <zodbot> dgilmore: mclasen
16:34:50 <paragan> I think I am okay with this change +1 but if there are questions good to get it discussed on list first
16:34:53 <dgilmore> proposal #agreed defer until next week. giving time for more discussion on the list
16:35:02 * mclasen_ looks up
16:35:11 <dgilmore> paragan: I honestly thing everyone is meh about it
16:35:28 <dgilmore> mclasen_: whats your thoughts on the pkgconfig change?
16:35:59 <dgilmore> mclasen_: https://fedoraproject.org/wiki/Changes/pkgconf_as_system_pkg-config_implementation is the change page
16:36:22 <mclasen_> I tried to stay clear of it - in my opinion pkg-config development was done loong ago, and it went downhill when somebody decided to make it circularly depend on glib to remove a few copied files
16:36:33 <mclasen_> and I think a full rewrite is entirelyt pointless
16:36:43 <mclasen_> but, I try not to care. As long as it doesn't break my builds
16:37:31 <dgilmore> okay
16:38:19 <ignatenkobrain> .hello ignatenkobrain
16:38:20 <zodbot> ignatenkobrain: ignatenkobrain 'Igor Gnatenko' <ignatenko@redhat.com>
16:38:30 <dgilmore> hey ignatenkobrain we were talking about https://fedoraproject.org/wiki/Changes/pkgconf_as_system_pkg-config_implementation
16:38:37 <dgilmore> everyone seems pretty meh on it
16:38:44 <ignatenkobrain> meh
16:38:47 <dgilmore> with no great opinion
16:38:59 <nirik> well, we are at +4 right?
16:39:14 <praiskup> I would be curious how many FTBFSs would happen in autotooled packages ...
16:39:15 <dgilmore> the consensus seems to be so long as nothing breaks then sure
16:39:23 <ignatenkobrain> praiskup: 0
16:39:31 <ignatenkobrain> pkg.m4 is 100% compatible
16:39:36 <ignatenkobrain> AFAICS
16:39:47 <ignatenkobrain> plan is to break 0 packages
16:39:50 <praiskup> ignatenkobrain, any attempts to "mass" build in copr on something like this?
16:40:18 <praiskup> s/on/or/
16:40:31 <ignatenkobrain> praiskup: I tried to build couple of packages locally in mock and it worked out
16:40:43 <nirik> I guess the thought is that we change it, mass rebuild and if it causes a lot of problems we could put the old one back and rebuild those failures.
16:41:09 <ignatenkobrain> nirik: that's the plan, if we can't fix things immediately - just postpone change
16:41:35 <ignatenkobrain> that's why I asked dgilmore about mass rebuild process that I would track it and fix issues related to change (if any)
16:42:12 <ignatenkobrain> also upstream developer is connected
16:42:19 <ignatenkobrain> so he'll fix our issues
16:42:24 <ignatenkobrain> kaniini on #fedora-devel
16:42:38 <dgilmore> I guess I am +1
16:42:41 <praiskup> ignatenkobrain, what distros has already done this change (except for *BSDs)?
16:42:56 <kalev> I'd be more comfortable with having this in immediately after we branch F27 so that there's time to see what breaks before doing a full mass rebuild
16:43:24 <sgallagh> kalev: Except that any breakage would really only be detectable by a mass-rebuild
16:43:25 <kalev> and then by that time the package will have been in Fedora for a bit as well and people have had time to play with it
16:43:28 <ignatenkobrain> kalev: thing is that we need mass rebuild
16:43:30 <sgallagh> So what would be the point?
16:43:46 <ignatenkobrain> to ensure that everything uses pkgconf
16:43:54 <ignatenkobrain> otherwise there can be some really weird bugs
16:43:58 <sgallagh> The chances of a random build hitting a bug are likely small
16:43:59 <ignatenkobrain> praiskup: not I'm aware of
16:44:22 <praiskup> ignatenkobrain, is GPL pkg-config dead?
16:44:29 <dgilmore> so we are at +5 now
16:44:33 <dgilmore> which is enough
16:44:47 <ignatenkobrain> praiskup: tl;dr it doesn't conform even their (freedesktop) specification
16:45:25 <praiskup> ignatenkobrain, then one should fix that ... not rewrite the software (the motivation behind rewrite sounds == license, or?)
16:45:36 <ignatenkobrain> praiskup: mainly bootstrapping issue
16:45:42 <ignatenkobrain> glib2 -> pkgconfig -> glib2
16:45:43 <kalev> sgallagh: I don't understand -- why is it neccessary to have a mass rebuild to see what breaks?
16:46:01 <ignatenkobrain> praiskup: check devel@
16:46:05 <ignatenkobrain> let me find message
16:46:12 <sgallagh> kalev: Because it's the only way to exercise the full set of uses in Fedora.
16:46:12 <kalev> sgallagh: as in, of course a mass rebuild would be an exhaustive test, but surely it's possibly to see how well it's doing _without_ a full mass rebuild?
16:46:21 <kalev> with just incremental builds going on in rawhide?
16:46:33 <ignatenkobrain> kalev: it would not have proper effect
16:46:38 <ignatenkobrain> and there can be some bugs
16:46:38 <dgilmore> #accepted F26 System Wide Change: pkgconf as system pkg-config implementation is accepted (5:+ 0:0 0:-)
16:46:43 <ignatenkobrain> if we do it invcrementally
16:46:45 <sgallagh> kalev: Because if anything breaks, it's likely to be the ancient stuff that's still in the distro because it never FTBFSes at the mass rebuilds
16:46:58 <sgallagh> Rather than the stuff that is actively maintained
16:47:02 <ignatenkobrain> praiskup: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/C7JZH3SJWDCACSOAYMXHJCT43OLWRJHD/
16:47:10 <sgallagh> /me speaks from recent Base Runtime experience
16:47:17 <sgallagh> Many of the build failures were due to bitrot.
16:47:20 <ignatenkobrain> hopefully this message will explain many of questions
16:47:37 <dgilmore> lets move on
16:47:42 <praiskup> good luck
16:47:46 <ignatenkobrain> dgilmore: thanks for invitation
16:48:30 <dgilmore> #topic #1635 F26 Self Contained Changes
16:48:38 <dgilmore> https://pagure.io/fesco/issue/1635
16:48:44 <dgilmore> .fesco 1635
16:48:45 <zodbot> dgilmore: Issue #1635: F26 Self Contained Changes - fesco - Pagure - https://pagure.io/fesco/issue/1635
16:49:18 <dgilmore> there is base runtime, in addition to the two we defered last week
16:50:17 <jsmith> I'm +1 for Base Runtime
16:50:37 <kalev> I'm +1 as well
16:50:38 <sgallagh> Did we get any more information from those two last week?
16:50:45 <paragan> +1
16:51:04 <nirik> +1 on base runtime
16:51:10 <jsmith> And frankly, I'm tired of waiting for more info on the other two, so consider me +1 for golang PIE and +1 for fontconfig cachey/indexy thingy
16:51:10 <sgallagh> I'm +1 on Base Runtime, with the notable bias of being one of the people working on it.
16:52:05 <sgallagh> I'd prefer to get a more specific request from the fontconfig people on what they want us to decide.
16:52:09 <sgallagh> As for golang... +1
16:52:10 <dgilmore> I am +1 to the base runtime modul;e
16:52:21 <dgilmore> the other two I do not rember seeing anything more
16:53:03 <dgilmore> paragan: what are you +1 on?
16:53:11 <sgallagh> That's partly my fault; I was supposed to talk with Colin about fontconfig and then forgot. Sorry about that.
16:53:29 <nirik> I guess I am +1 on the other two. But I'd be ok with more info on the font thing also
16:53:40 <paragan> +1 for base runtime
16:53:45 <dgilmore> #action sgallagh to talk to walters about fontconfig
16:54:39 <paragan> +1 for golang also
16:54:47 <dgilmore> #agreed base runtime is accepted (6:1 0:0 0:-)
16:54:47 <kalev> +1 from golang as well
16:55:23 <jsmith> I think there's plenty of information in the bz issue for the fontconfig thing
16:55:24 <dgilmore> +1 to golang any issues I am sure will get addressed
16:55:33 <jsmith> Certainly enough there to change my mind :-)
16:57:10 <dgilmore> we are +4 for golang
16:57:53 <dgilmore> and +1 on fontconfig
16:57:57 <sgallagh> dgilmore: I count +5 on golang
16:58:44 <dgilmore> sgallagh: who?
16:58:54 <sgallagh> me, nirik, paragan, kalev, you
16:58:57 <sgallagh> In that order
16:59:06 <jsmith> What about my votes?
16:59:26 <jsmith> Doesn't that make +6?
16:59:27 <sgallagh> jsmith: I don't see a vote from you on golang
16:59:39 <sgallagh> Oh, wait
16:59:45 <sgallagh> Sorry, two lines above mine. Mea culpa
16:59:45 <dgilmore> jsmith: I was looking too close to the left
17:00:03 <dgilmore> #agreed golang PIE change is accepted (6:1 0:0 0:-)
17:00:21 <dgilmore> so that leaves fontconfig
17:00:31 <dgilmore> I have jsmith as +1
17:00:38 <dgilmore> I can go +1 also
17:00:46 <sgallagh> dgilmore: I just emailed Colin and will meet with him next week. I'm +0 until I do.
17:01:30 <kalev> sure, let's defer then, but generally I think it's fine to have those kinds rpm regenerated caches in /usr
17:01:48 <kalev> would be good to make sure that this proposal makes sense to colin though
17:02:40 <sgallagh> kalev: I don't agree, actually. Which is why I'm going to discuss it with Colin :)
17:03:14 <kalev> generally I think things in /var/cache should be those kinds of caches that can be regenerated at runtime
17:03:39 <sgallagh> kalev: tl;dr version: IMHO /var/cache is very wrong, /usr is only *less* wrong.
17:03:49 <kalev> but the fontconfig one cannot, as it's regenerated with rpm scripts, so I think it makes sense to have it somewhere else
17:03:52 <dgilmore> proposal #agreed defer fonconfig until after sgallagh talks to walters
17:03:58 <kalev> yup, agreed with that sgallagh
17:04:14 <sgallagh> /me takes it out of the meeting
17:04:15 <kalev> /var/lib maybe then? anyway, Colin is the expert there :)
17:04:18 <kalev> +1 to defer
17:04:41 <paragan> +1 to defer
17:05:13 <dgilmore> +1 to defering
17:05:43 <sgallagh> +1 defer
17:05:59 <dgilmore> nirik: jsmith: ?
17:06:11 <nirik> sure. +1 defer
17:06:40 <dgilmore> #agreed defer fonconfig until after sgallagh talks to walters (5:+ 0:0 0:-)
17:07:00 <dgilmore> #topic Next week's chair
17:07:17 <dgilmore> who would like to run next week?
17:08:01 <nirik> I've not done it in a while, I can.
17:08:11 <kalev> I haven't done it in a while, but I'm also up for reelection, so not sure I should volunteer ... :)
17:08:17 <sgallagh> /me was planning to drink himself unconscious on Inauguration Day, frankly :-/
17:08:20 * nirik also
17:08:28 <dgilmore> #action nirik to run next week
17:08:32 <dgilmore> thanks nirik
17:08:43 <dgilmore> #topic Open Floor
17:08:45 <nirik> should we get someone not up for relection?
17:08:57 <dgilmore> nirik: possibly
17:09:05 <nirik> that will be the next fesco... election ends on the 16th...
17:09:06 <dgilmore> #undo
17:09:06 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x41b06410>
17:09:07 <sgallagh> If nirik isn't re-elected, I'll take it
17:09:28 <dgilmore> #info sgallagh will run the meeting if nirik is not reelected
17:09:39 <nirik> ok.
17:09:46 <dgilmore> #topic Open Floor
17:09:54 <dgilmore> okay does anyone have anything?
17:10:43 <nirik> Everyone should remember to go vote in fesco/famsco/council elections...
17:10:56 <dgilmore> #info 17:10 < nirik> Everyone should remember to go vote in fesco/famsco/council elections...
17:11:15 <sgallagh> I voted! (Where's my sticker?)
17:11:18 <dgilmore> Iw ill end in a minute if nothing else
17:12:14 * paragan also voted already
17:14:47 <dgilmore> #endmeeting