17:00:11 <geppetto> #startmeeting fpc
17:00:11 <zodbot> Meeting started Thu Dec  2 17:00:11 2021 UTC.
17:00:11 <zodbot> This meeting is logged and archived in a public location.
17:00:11 <zodbot> The chair is geppetto. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
17:00:11 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:00:11 <zodbot> The meeting name has been set to 'fpc'
17:00:11 <geppetto> #meetingname fpc
17:00:11 <zodbot> The meeting name has been set to 'fpc'
17:00:11 <geppetto> #topic Roll Call
17:00:17 * GwynCieslasheher here
17:00:28 <geppetto> #chair GwynCieslasheher
17:00:28 <zodbot> Current chairs: GwynCieslasheher geppetto
17:01:46 <tibbs> Hey
17:01:52 <geppetto> #chair tibbs
17:01:52 <zodbot> Current chairs: GwynCieslasheher geppetto tibbs
17:02:04 <tibbs> Didn't realize we were on for today.
17:02:40 <geppetto> I don't think it's 100% even now :)
17:04:43 <tibbs> And imediately after I typed that, my X server died.
17:05:43 <GwynCieslasheher> Nice
17:09:26 <tibbs> There has been a bit of new activity but I haven't had a time to look over it.
17:09:50 <geppetto> #topic Open Floor
17:09:58 <geppetto> Yeh, the only real thing I had was: https://pagure.io/packaging-committee/issue/1132
17:10:12 <geppetto> Which confused me at least as much as it confused tibbs
17:10:26 * GwynCieslasheher reads
17:11:01 <tibbs> I think what they want is something that can be read and used by automation.
17:11:44 <GwynCieslasheher> I think I grok what they want but it seems more complex than what we currently do with scripts.
17:11:54 <tibbs> I wouldn't really object to actually recommending that a shell script be used if possible, and that it have a fixed name.
17:12:08 <GwynCieslasheher> That's a good idea.
17:13:03 <tibbs> But I don't know how far you go with it.  Because they still wouldn't know how to get the unmodified source.
17:13:46 <GwynCieslasheher> Mandate a comment in the script with a URL?
17:13:59 <tibbs> Yeah, that's all I can think of.
17:14:13 <tibbs> Unmodified source: https://whatever.tgz
17:14:28 <tibbs> I'll add those thoughts to the ticket
17:14:50 <geppetto> Yeh, I'm +1 on comments for people … but comments for automation seem like a bad idea
17:15:15 <tibbs> Sadly you can't reference it with a Source: tag.
17:15:19 <tibbs> NoSource, maybe?
17:15:37 <tibbs> Do those still exist?
17:16:36 <tibbs> Seems to, but that may be a bit out there.
17:18:07 <GwynCieslasheher> I'd want to think long and hard about that, so we don't push packagers from something was simple as a bicycle onto what might turn out to be an electrified penny farthing with a spoiler but no tires.
17:18:38 <GwynCieslasheher> So to speak.
17:19:03 <tibbs> Right, that's the problem.  Making it suitable for automation is great, but that might just add complexity and push it onto the packager.
17:19:11 <geppetto> yeh, I think I'm -1 on using nosource for things that have source generated
17:19:43 <geppetto> like if they want automation things I'd heavily prefer them to speak to rpm people about adding a new tag
17:20:01 <GwynCieslasheher> Thinking further, if we had that, the build process would require network access beyond SCM and lookaside. Which changes tons of assumptions.
17:20:08 <geppetto> and I realize that might be a black hole of nothing … but I still can't think of a better solution
17:20:23 <tibbs> The problem with nosource is that if you leave it out then you end up with stuff in the srpm that shouldn't be there.
17:20:37 <GwynCieslasheher> Bingo. Or requiring net access.
17:20:51 <GwynCieslasheher> The first is a hard no, the second something of a PITA.
17:20:56 <tibbs> Because it you don't do NoSource: https://whatever.tgz, you have a regular Source: URL and then say "don't put it in the SRPM" with NoSource: 2 or whatever.
17:21:11 <geppetto> Yeh, I think maven kind of works like that for java things?
17:21:17 <tibbs> I don't see how any of this would require network access, though.
17:21:34 <tibbs> You have to run this automation before you can even generate the SRPM.
17:21:36 <geppetto> So it's possible … just a giant PITA
17:22:07 <GwynCieslasheher> Ah, so the packager runs the automation and the products go in lookaside just like now.
17:22:16 <GwynCieslasheher> Yeah.
17:22:20 <GwynCieslasheher> Ick
17:24:57 <decathorpe> hey o/ sorry for being late, I didn't get IRC notifications for some reason, and I didn't know there'd be a meeting today
17:24:59 <tibbs> I mean, it would be simpler for everyone (except the person who has to initially implement the thing) if there was one script that took a version, downloaded the unshippable tarball and generated the shippable one.
17:25:51 <tibbs> That's what I have done in the past, because it makes maintenance super easy.  But mandating that for all packages that need to do this seems to be too much, even if there aren't that many.
17:27:34 <GwynCieslasheher> RIght. And it can take the new version number as an argument.
17:27:49 <geppetto> yeh, certainly sometimes you can't easily get any version … only the latest etc.
17:28:01 <geppetto> #chair decathorpe
17:28:01 <zodbot> Current chairs: GwynCieslasheher decathorpe geppetto tibbs
17:28:39 <tibbs> I just think that doing it in a way that can be completely automated gets far more complicated than anyone would really want.
17:28:47 <geppetto> But, again, I'm happy to +1 almost any best practice on comments … but much less so on anything that should be read by automation
17:28:59 <GwynCieslasheher> Agreed
17:29:09 <geppetto> So depending on what they want is going to depend on how hard they need to work to convince me
17:29:19 <tibbs> But maybe the people who would want this would have better ideas.  It just feels to me like it wasn't really thought through all the way.
17:29:49 <geppetto> ¯\_(ツ)_/¯
17:30:17 <geppetto> Anyway … we've been here 30 mins. … so I feel like I shoud give everyone 30 mins. back
17:30:39 <tibbs> I can use that to finish typing this comment in the ticket.
17:30:42 <geppetto> I should find time to do an email schedule etc. for next week, so hopefully more people will realize they'll be a meeting :)
17:30:48 * geppetto nods
17:32:01 <decathorpe> geppetto: I realize sending out the email is more work, but it gives people more than 5 minutes of prior notice ;)
17:32:34 <geppetto> I know, I've just been busy and let it slide
17:33:02 <tibbs> We're all busy; it's no big deal.
17:33:15 <GwynCieslasheher> I have Element set to notify on #meetingname fpc
17:33:16 <geppetto> 100% my fault, but should be fixed for next week.
17:34:08 <tibbs> You know, I wonder if fedpkg could somehow become responsible for generating safe versions of unshippable tarballs.
17:34:20 <geppetto> lol
17:34:38 <GwynCieslasheher> That would be a logical spot for it but I'm not volunteering to write that code.
17:35:14 <tibbs> Maybe rephrase as asking the people who want this to leverage fedpkg.
17:35:45 <tibbs> If fedpkg requires some kind of metadata or helper scripts or whatever, it would be outside of the specfile and Not Our Problem (TM).
17:35:48 <geppetto> If they can get something into fedpkg I'm happy with it … if only because it's not my problem ;)
17:36:04 * geppetto hi5 tibbs
17:36:06 <tibbs> We'd merely document what fedpkg needs and be happy.
17:36:24 <geppetto> And on that note...
17:36:27 <geppetto> #endmeeting