16:00:06 <geppetto> #startmeeting fpc
16:00:06 <zodbot> Meeting started Thu Jul  9 16:00:06 2015 UTC.  The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:06 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:06 <geppetto> #meetingname fpc
16:00:06 <zodbot> The meeting name has been set to 'fpc'
16:00:06 <geppetto> #topic Roll Call
16:00:10 <tibbs|w> Howdy.
16:00:13 <tomspur> hi
16:00:34 <geppetto> #chair tomspur
16:00:34 <zodbot> Current chairs: geppetto tomspur
16:00:36 <geppetto> #chair tibbs
16:00:36 <zodbot> Current chairs: geppetto tibbs tomspur
16:00:59 * tomspur won't be here next week
16:01:14 <geppetto> ok
16:01:19 <tomspur> Should I send a mail to the list or is an entry in the vacation calender enough?
16:01:20 <geppetto> #chair orionp
16:01:20 <zodbot> Current chairs: geppetto orionp tibbs tomspur
16:01:34 <orionp> morning
16:01:51 <geppetto> tomspur: If you can reply to next weeks schedule, that's easiest … if you won't be near a computer the day before, then it doesn't matter … I'll probably remember :)
16:02:07 <tomspur> geppetto: ok :)
16:02:28 <tibbs|w> If we knew in advance that we wouldn't get quorum we could just postpone the meeting.
16:03:01 <tibbs|w> Would be super nice if everyone who couldn't make it would tell us that they couldn't make it.
16:03:11 <geppetto> yeh
16:03:44 <tibbs|w> We did have that nice stretch where we had plenty of people, but not much luck lately.
16:04:38 <geppetto> yeh
16:04:42 <geppetto> it is summer though
16:04:46 <geppetto> outside is nice and warm
16:04:56 <tibbs|w> Depends on where you are.
16:04:59 <geppetto> ha
16:05:03 <orionp> not here - cold and rainy
16:05:06 <tibbs|w> This is the time of year when we don't go outside.
16:05:28 <orionp> although that's not typical here
16:06:06 <tibbs|w> Right now it's only 31C but with 99.99999% humidity.
16:06:06 <orionp> I think my TB calendar reminders are back to hiding again
16:06:37 <tibbs|w> I'm doing some more cleanup on the tickets and then I just need to write up all of the python stuff and I'll actually be caught up.
16:07:13 <tomspur> I cannot get python3 built on f22 otherwise I'd start with the updates to ship the macros... :/
16:07:36 <geppetto> why doesn't it build?
16:07:37 <tibbs|w> I don't know whether to close the Go ticket or to keep bugging them.
16:07:42 <tomspur> Will ask the python folks to just disable the one failing test...
16:07:46 <tibbs|w> We do need guidelines but I sure as hell can't write them.
16:08:34 <tomspur> geppetto: one failing test in the testsuite. test_gdb is disabled on a lot of arches and maybe just count arm into it...
16:08:45 * geppetto nods
16:10:02 <tibbs|w> So....
16:10:12 <geppetto> Yeh, I don't think we are going to get anymore people
16:10:26 <geppetto> I've tried to ping the other 4 (mbooth said he couldn't make it)
16:10:30 <geppetto> And nobody is on IRC
16:10:44 <geppetto> We could talk about the bundling ticket … but I'm not sure it's worth it
16:11:05 <tibbs|w> Any opinions on what I should do with the Go ticket?
16:11:18 <tibbs|w> https://fedorahosted.org/fpc/ticket/382
16:11:27 <geppetto> I'm not sure
16:11:44 <tibbs|w> It's been something like four months since there was any information there about the actual guidelines.
16:12:36 <tomspur> hmm, maybe ask on the devel list if there are some people interested in it and want to continue?
16:13:05 <tibbs|w> I added something.
16:13:34 <tibbs|w> Honestly I think we shouldn't have had a single Go package in the distribution without at least minimal guidelines.
16:13:38 <tibbs|w> But reviewers don't care.
16:16:20 <geppetto> yeh, I mean some might
16:16:35 <geppetto> But esp. with docker/etc. … there's a lot of push to get those in
16:16:51 <geppetto> And Fedora isn't traditionally good at saying no.
16:16:58 <tibbs|w> Then there should have been a lot of push to get guidelines in.
16:17:10 <geppetto> #chair Rathann
16:17:10 <zodbot> Current chairs: Rathann geppetto orionp tibbs tomspur
16:17:14 <Rathann> hello, sorry for being late
16:17:16 <geppetto> And then there were 5!
16:17:22 * Rathann is on vacation
16:17:26 <geppetto> Ahh
16:17:28 <tibbs|w> It wouldn't have been hard to have something minimal.  "This is where your files go.  Here's an example spec.  Suggest using this utility to generate packaging.".
16:17:32 <geppetto> #topic Schedule
16:17:34 <geppetto> https://lists.fedoraproject.org/pipermail/packaging/2015-July/010808.html
16:17:53 <geppetto> #topic #550 	Darktable and Rawspeed internal library
16:17:53 <geppetto> .fpc 550
16:17:53 <geppetto> https://fedorahosted.org/fpc/ticket/550
16:17:54 <zodbot> geppetto: #550 (Darktable and Rawspeed internal library) – fpc - https://fedorahosted.org/fpc/ticket/550
16:18:17 <geppetto> tibbs: Yeh, if we are going to just bless some minimal thing it might be easy to get someone to write that
16:18:39 <geppetto> tibbs: The main problem was that all of it was a giant static lib. joke.
16:18:45 <geppetto> IIRC
16:19:01 <tibbs|w> That's true, but if it was that bad the packages shouldn't be in there at all.
16:19:17 <tibbs|w> And if it's OK for the packages to be in there then there should at least be guidelines.
16:19:22 <geppetto> yeh, I agree, but see about about Fedora and saying no.
16:20:02 <geppetto> I think it's a giant mess waiting for a disaster to happen … but, meh.
16:20:19 <tomspur> "RawSpeed is not at the moment a separate library, so you have to include it in your project directly." It seems that's the only reason for this bundled library exception request
16:20:58 <tomspur> but sorry for interrupting :)
16:21:09 <geppetto> nah, it's fine … I did change the topic :)
16:21:26 <geppetto> Yeh, rawcode looks like it's fairly active … and huge.
16:21:34 <geppetto> rawspeed, even
16:22:11 * Rathann notes nobody filed a bug saying it should be buildable as a shared library
16:22:27 <tibbs|w> But pretty much anything can be built as some kind of library.
16:22:32 <Rathann> ah wait
16:22:33 <tibbs|w> Even static is an improvement.
16:22:36 <Rathann> https://github.com/klauspost/rawspeed/issues/109
16:22:43 <geppetto> wc -l *.cpp is like 30k
16:22:54 <geppetto> so lol on that being a copylib
16:23:31 <geppetto> another 10k in headers
16:24:07 <geppetto> cool the git repo. has static binaries in it
16:24:31 <tomspur> I don't understand what they mean with a "release" when reading the ticket above...
16:25:06 <tomspur> Users should to a git checkout and "support for their [...] releases"
16:25:38 <tomspur> if there are only new cameras added it doesn't sound like they would break anything else much
16:25:47 <orionp> This seems to be what we're coming too, just mash everything together
16:26:51 <geppetto> tomspur: seems like general bug fixes etc. are happening
16:27:07 <Rathann> looking at github graphs, it's actively developed
16:27:12 <orionp> So why call it "Raw Image Decoder *Library*"?
16:27:59 <Rathann> it's fine if they don't want to make releases or guarantee API/ABI stability, but the packager should explain to them that these things make users' and packagers' lives easier
16:28:19 <Rathann> so a static library is a minimum requirement in this case I'd say
16:28:57 <geppetto> yeh
16:29:08 <Rathann> "is extensively crash-tested on broken files." - this is encouraging, but doesn't mean there will be no security bugs
16:29:47 <Rathann> also, the readme says something about "Version 2"
16:29:48 <geppetto> yeh, esp. considering: https://github.com/klauspost/rawspeed/commit/ebfe0d82a89623dd04fdc3db2693a5b288175e8c
16:30:11 <tibbs|w> Yeah, sorry, there's just no way I'm going to go for this.
16:31:01 <geppetto> I assume you are fine with a static lib?
16:31:07 <Rathann> also, it looks like it bundles pugixml, which we have packaged
16:31:54 <tibbs|w> Yeah, static lib is less good but still better.
16:31:55 <geppetto> rawspeed or darktable?
16:32:10 <Rathann> rawspeed
16:32:15 <geppetto> ok
16:32:45 <geppetto> #action Rawspeed is way too big/active to be a copylib. Make a static library.
16:33:01 <Rathann> and educate upstream about the benefits of shared libraries
16:33:01 <geppetto> #action Also rawspeed needs to unbundle pugixml
16:33:02 <tibbs|w> But if it has to unbundle something, that forces it to be a shared lib.
16:33:15 <Rathann> and of making proper releases with API/ABI stabilit
16:33:16 <Rathann> y
16:33:18 <tibbs|w> Unless someone makes a pugixml static package.
16:33:35 <tibbs|w> We can't force upstream to make proper releases, though it would be nice.
16:33:43 <geppetto> yeh
16:34:00 <Rathann> I'm not saying force, but educate and if possible provide a patch
16:34:01 <geppetto> I'd like proper upstream releases and a shared lib. … also a unicorn ;)
16:34:04 <Rathann> hehe
16:34:13 <tibbs|w> I'd settle for a pony.
16:34:16 <tibbs|w> Or a Dr. Pepper.
16:34:20 <geppetto> :)
16:34:28 <Rathann> I believe upstream needs to grow a build system first
16:34:49 <tibbs|w> It would be nice, but not really relevant.
16:35:09 <geppetto> #topic #508 	New GID for openstack-neutron
16:35:09 <geppetto> .fpc 508
16:35:09 <geppetto> https://fedorahosted.org/fpc/ticket/508
16:35:10 <zodbot> geppetto: #508 (New GID for openstack-neutron) – fpc - https://fedorahosted.org/fpc/ticket/508
16:35:11 <tomspur> It seems upstream doesn't know why we want to unbundle. Maybe they at least do some (moving) tags from time to time :)
16:35:13 <tibbs|w> Fedora has to adapt and force it into some reasonable shape for packaging, even if upstream never takes it.
16:35:17 <geppetto> Pretty sure there's nothing to do here
16:35:26 <tibbs|w> Yeah, I think we should just close this.
16:35:48 <tibbs|w> But I do need to get back to that fun hack I was talking about in that ticket.
16:36:16 * geppetto nods
16:36:29 <geppetto> I'm not sure about comment 3
16:36:47 <geppetto> I think he's just saying "this would be nice … because cloud"
16:37:01 <geppetto> But maybe I'm missing something
16:37:05 <tibbs|w> I agree.
16:37:23 <tibbs|w> They don't say what breaks.
16:37:44 <geppetto> Or why this is different to any other random service that runs on cloud
16:37:48 <tibbs|w> There's really no argument at all besides "someone thinks they need it".  They haven't articulated why they need it.
16:38:02 <tibbs|w> Or how having a fixed gid will help anyone setup ldap.
16:38:28 <tibbs|w> Also, how did the openstack things get gids assigned in the 160-170 range?
16:38:43 <tibbs|w> I don't recall having that discussion, but there are plenty of things I don't recall these days.
16:39:13 <geppetto> 322 and 396 are the ones I can find
16:39:28 <geppetto> both of which are closed as negatives
16:40:12 <geppetto> so it looks like someone just allocated a bunch of ones they wanted
16:40:19 <tibbs|w> "someone".
16:40:54 <Rathann> https://bugzilla.redhat.com/show_bug.cgi?id=902987
16:41:03 <tibbs|w> I don't want to get into the "disparaging Red Hat" business but we all know how this stuff happens.
16:41:05 <Rathann> that's for 165
16:41:19 <tibbs|w> Nice.
16:41:35 <tibbs|w> We could save a lot of time by not caring.
16:41:40 <geppetto> ha
16:41:48 <Rathann> https://bugzilla.redhat.com/show_bug.cgi?id=845078
16:41:54 <Rathann> 187
16:41:57 <tibbs|w> If nobody else does, what's the point?
16:42:35 <geppetto> so … doesn't look like any of these openstack daemons need a static uid
16:42:39 <Rathann> https://bugzilla.redhat.com/show_bug.cgi?id=732442
16:43:09 <Rathann> 160-162
16:43:11 <tibbs|w> geppetto: I agree, but they gave them out anyway.  Maybe it's just not as big of a deal as we thought it was.
16:43:15 <geppetto> wow, the best reasoning yet
16:44:05 <geppetto> To be fair … did we always look at this?
16:44:07 <Rathann> either we should review all static uidgid assignments or none - a lot of them seem to have sneaked in via bugzilla
16:44:18 <Rathann> not always
16:44:31 <tibbs|w> No, we didn't always look at this.
16:44:37 * geppetto is trying to remember if this used to be just "whatever the setup maintainer felt like doing" before 2013ish
16:44:39 <tibbs|w> These days I follow commits to the setup package.
16:44:50 * geppetto nods
16:44:59 <tibbs|w> But fedmsg has really made it difficult to follow things now, unless you like mail.
16:45:22 <geppetto> You can't filter things?
16:45:29 <tibbs|w> If you can figure out the interface.
16:45:33 <geppetto> ahh
16:45:53 <tibbs|w> There's no real way to do "this is what I want to see".
16:46:27 <tomspur> Hmm, what do you try to filter? It took me a while, but seems working (mostly)
16:46:31 <tibbs|w> Essentially all of my mail is ABRT reports, and setting the ABRT filter either way still gets me the reports.  I had just clicked "watchcommits" on the kernel package.
16:46:38 <tibbs|w> Anyway, offtopic.
16:47:24 <tibbs|w> I'm still not in favor of the GID thing, and think that the other openstack ones shouldn't have been granted, either.
16:47:31 <geppetto> Kind of getting back to just rubber stamping uid allocation requests
16:47:52 <geppetto> Is there anyone we want to ask if we should be a lot more rubber stampy about these?
16:48:00 <geppetto> fesco?
16:48:02 <geppetto> fpl?
16:48:03 <tibbs|w> FESCo, I guess.
16:48:33 <tibbs|w> I remember hearing someone say that our pool was large enough to accommodate a whole mess of allocations.
16:48:48 <tibbs|w> If that's the case, then no big deal.  But I thought we had only tens left.
16:48:48 <geppetto> yeh
16:49:02 <geppetto> I thought we'd got a lot more than that
16:49:08 <geppetto> but I wouldn't bet a lot on it
16:49:14 * Corey84 listens from back row
16:49:23 <tibbs|w> Maybe someone decided to say "you have a user with uid 500?  too damn bad."
16:49:32 <geppetto> The main thing was that we can't ever really get any back … so we should be conservative in allocating anyway
16:49:50 <geppetto> tibbs: Yeh, default is now 1000 for first user, right?
16:50:01 <geppetto> Not sure what happens when/if we hit 500 though
16:50:07 <tibbs|w> Yes, it's 1000 now.
16:50:11 <tibbs|w> I don't remember when that changed.
16:50:26 <geppetto> a lot of people are going to find out what backwards compat. means at that point I guess
16:50:40 <geppetto> A few years ago
16:50:40 <tibbs|w> RHEL5 has it at 500.
16:50:58 <tibbs|w> RHEL6 has it at 500.
16:51:06 <geppetto> ugh
16:51:14 <tibbs|w> RHEL7 has it at 1000.
16:51:15 <geppetto> would have bet el6 had 1000
16:51:24 <geppetto> so … super super recent then
16:52:51 <tibbs|w> And of course in that 500 we still have to squeeze the static pool as well as the dynamic pool.
16:53:28 <Rathann> so... getting back to #508, I'm -1 unless they provide some specific scenario where having in dynamically allocated breaks things
16:53:49 <Rathann> s/in/it/
16:53:51 <tibbs|w> Yeah, they should always describe in detail what breaks.
16:54:08 <tibbs|w> "We need it because cloud" is not really sufficient.
16:54:30 <tibbs|w> "We need it because it makes some unspecified or unrelated thing easier" isn't either.
16:54:36 <geppetto> I think the problem is that nothing breaks
16:57:34 <geppetto> #action Looking at previous openstack uid/gids we wouldn't have approved any of them, so look at filing a FESCo ticket to see if we should pass a lot more of these that don't really need it.
16:57:39 <geppetto> #topic Open Floor
16:58:02 <geppetto> Anyone want to volunteer to open the fesco ticket ?;)
16:58:12 <tibbs|w> I have nothing.  Still hoping that a python with the new macros makes it to F22 soon.
16:58:41 <geppetto> yeh, that's on tomspur … right? Should happen when the py3 stuff gets solved?
16:59:18 <tibbs|w> It is coming, so I think I'll just push the new guidelines over the next couple of days.
16:59:33 * geppetto nods
16:59:36 <tibbs|w> There's always going to be confusion when we change this much anyway.
16:59:43 <geppetto> yeh
16:59:54 <tomspur> geppetto: yes. I just pushed the python2 to testing just in case
17:00:07 <geppetto> Can obviously tweak some stuff if anything comes to light anyway
17:00:07 <tomspur> Will try to ping the python folks tomorrow if it is save to just disable the test
17:00:08 * geppetto nods
17:00:18 <tomspur> I don't think I should just disable it without them knowing
17:00:52 <tibbs|w> You can always turn it back on.
17:01:14 <tibbs|w> But, sure, telling people is never bad.
17:01:19 <geppetto> yeh, the way I've done stuff like that in my packages if have a %global variables that controls it
17:01:35 <geppetto> Then it's easy to turn off if upstream fixes the test/whatever
17:01:50 <tibbs|w> So we're really getting the ticket count down.
17:02:08 <tibbs|w> I have one procedure to write up and three python tickets.
17:02:13 <tomspur> I just needed to backport a fix to get it built with a newer openssl, so python was FTBFS right now...
17:03:03 <tibbs|w> I don't think anything will ever happen with the scintilla thing.  I should file some bugs but I don't expect anything to happen.
17:03:22 * geppetto nods
17:03:29 <tibbs|w> And I think that kind of completes everything we have going that isn't needinfo.
17:03:43 <geppetto> yeh
17:03:48 <Rathann> well, at the very least they should add provides: bundled(scintilla)
17:03:51 <geppetto> report 12 and 13 are pretty much empty
17:03:54 <Rathann> I think we requested that
17:04:07 <tibbs|w> Rathann: I actually did that.
17:04:10 <Rathann> cool
17:04:59 <tibbs|w> Not really much we can do unless someone wants to maintain an actual scintilla package.
17:05:15 <tibbs|w> (And I sure don't want to do it by myself.)
17:05:19 <geppetto> So … s/unless.*//
17:05:35 <tibbs|w> rdieter had talked about it.
17:05:42 <geppetto> interesting
17:05:45 <tibbs|w> But he has enough to do.
17:06:49 <geppetto> yeh, just surprised someone was thinking about it
17:07:05 <tibbs|w> He said he was just waiting for a bit of free time, but that was a couple of months ago.
17:07:26 <geppetto> Ha, yeh I know how that goes
17:08:43 <geppetto> Ok, see everyone but tomspur next week :)
17:08:48 <geppetto> thanks for coming
17:08:53 <tibbs|w> Thanks for running.
17:09:02 <Rathann> thank you
17:09:44 <tibbs|w> tomspur: Do let me know if you happen to get the updated py3 out to F22.
17:10:35 <geppetto> #endmeeting