16:00:26 #startmeeting fpc 16:00:26 Meeting started Thu Aug 20 16:00:26 2020 UTC. 16:00:26 This meeting is logged and archived in a public location. 16:00:26 The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:26 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00:26 The meeting name has been set to 'fpc' 16:00:26 #meetingname fpc 16:00:26 #topic Roll Call 16:00:26 The meeting name has been set to 'fpc' 16:00:35 * limburgher here 16:00:47 #chair limburgher 16:00:47 Current chairs: geppetto limburgher 16:00:51 .hello ngompa 16:00:52 King_InuYasha: ngompa 'Neal Gompa' 16:01:23 Hey, Neal … is that nick from an anime? 16:01:29 .hello2 16:01:30 decathorpe: decathorpe 'Fabio Valentini' 16:01:36 #chair decathorpe 16:01:36 Current chairs: decathorpe geppetto limburgher 16:02:55 .hello2 16:02:56 defolos: defolos 'Dan Čermák' 16:03:44 hey 16:04:06 #chair mhroncok 16:04:06 Current chairs: decathorpe geppetto limburgher mhroncok 16:04:07 Hey 16:04:28 Very programing atm … one off error for quorum ;) 16:05:01 ±1 depending on how we count? :) 16:05:34 Hey, sorry. 16:06:17 #chair tibbs 16:06:17 Current chairs: decathorpe geppetto limburgher mhroncok tibbs 16:06:25 Het, better late than never ;) 16:06:35 And we have 5 now 16:06:43 :partyparrot: 16:09:28 .hello2 16:09:29 carlwgeorge: carlwgeorge 'None' 16:09:37 geppetto: yep 16:09:46 InuYasha, of course ;) 16:10:17 #topic Schedule 16:10:20 #link https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/DWIAFVVQVOFHJLSHJ4T6GUYUE5JPDPUH/ 16:10:40 King_InuYasha: Cool, I wondered … as I've seen a few episodes. But wasn't sure :) 16:11:00 it's a good show, though has a weak beginning 16:11:17 defolos: Hey, so did you turn up for s specific issue/PR/etc? 16:12:07 geppetto: I was hoping I could "encourage" Igor Raits to work further on adding the list of packages to the mock build root 16:12:50 so that I could find out which version of a package is currently in the buildroot without having to ask rpm directly, because that is apparently not a good idea 16:13:01 Ahh 16:13:43 and if that list contained all the bconds, that would be even better 16:13:46 #topic #pr-954 Prohibit use of `rpm` command from specfile. 16:13:47 https://pagure.io/packaging-committee/pull-request/954 16:13:59 I disagree with the premise that we should block it 16:14:04 then I could have even more automated bootstrapping 16:14:15 I too 😉 16:14:25 and absent an alternative, we shouldn't even bother with this 16:14:29 but I really don't know enough about this to have a qualified opinion on this 16:14:30 King_InuYasha: You think people should call rpm from mock/rpmbuild? 16:14:37 why not? 16:15:01 King_InuYasha: I really don't like it when .spec file evaluation runs arbitrary code on my system :( 16:15:06 the only reason it *might* not work is that we don't have Koji running in bootstrap mode 16:15:19 err running builds in bootstrap mode 16:15:20 which we should enable :) 16:15:25 I think Panu mentioned that with bootstrap chroot it's not a problem, but neither COPR nor koji do that, yeah 16:15:33 COPR does use bootstrap chroot 16:15:39 and Mock itself does by default too 16:15:42 only Koji does not yet 16:16:06 COPR does *NOT* by default, but you can enable it "experimentally" 16:16:14 King_InuYasha: https://pagure.io/packaging-committee/pull-request/954#comment-112881 ? 16:16:23 huh, that's odd, I recall FrostyX telling me he was going to switch it 16:16:42 There's a settings switch with this text: "Enable mock's use_bootstrap_container experimental feature" 16:16:43 We should at least have a 'Please don't', to avoid rpm recursion black holes... 16:17:09 decathorpe: he was going to switch it to default to that behavior a while ago because of problems evaluating macros in spec files across chroots 16:17:20 e.g. Mageia's %mkrel doesn't exist in Fedora 16:17:32 so Mageia spec files don't work without bootstrap mode 16:17:49 and openSUSE's %python_subpackages macro doesn't work for the same reason 16:18:15 huh. a COPR I created two days ago doesn't have it enabled 16:18:33 #chair mhroncok_mobile 16:18:33 Current chairs: decathorpe geppetto limburgher mhroncok mhroncok_mobile tibbs 16:18:35 (%python_subpackages macro generates subpackages for various Python versions enabled out of the box by openSUSE for packaging, so they generate flavor builds) 16:19:02 e.g. https://build.opensuse.org/package/view_file/devel:languages:python/python-iniparse/python-iniparse.spec?expand=1 16:19:35 stuff like this simply fails without bootstrap chroot *anyway* 16:19:55 so koji should enable bootstrap chroot is what you're saying? :) 16:19:57 yes 16:20:00 Not sure what relevance these examples from other distros actually have. 16:20:15 tibbs: we have variations of this in Fedora and RHEL too 16:20:24 with %py3_dist macros for BRs and such 16:20:29 Maybe those things are just broken or are too fancy for their own good and make assumptions that aren't valid. 16:20:39 sorry, I don't buy it 16:20:51 Which, since they don't work, is possible. 16:21:21 in *general* spec files are written to be evaluated on the target distro 16:21:34 the fact that Koji violates this rule pretty flagrantly is a problem 16:21:39 Remember that we work to document the interaction of the tools that we have. If it doesn't work in koji then we don't tell people to use it. 16:21:57 Our interest here is not things from other distro, it is Fedora. 16:22:14 well, again, I can come up with Fedora/EPEL cases too 16:22:31 You mean things that people do which they shouldn't be doing because they don't work? 16:22:40 Saying that if mock will change to use bootstrap mode by default, and koji goes with that too … then I'm fine with saying people can use things that work in that mode 16:22:52 geppetto: I'm happy with that too 16:23:13 and fwiw, koji *will* switch to this eventually 16:23:23 Can we define eventaully? 16:23:24 I mean, we can't put that in a packaging guideline. "It doesn't work, but you could use it if somehow it starts working". 16:23:44 geppetto: last I discussed it with them, probably within the next six months or so 16:23:59 tibbs: I was thinking more like … this should start working with F-33 or F-34 … please don't use it before then, as it might break 16:24:05 once Fedora 34 drops bdb support, it needs to be in place 16:24:12 (bdb rpmdb, that is) 16:24:14 Best to either put the whole thing on ice, or if it's going to be too long, document that it doesn't work and then change that when the time is right. 16:25:03 tibbs: I agree, dealing with this now seems premature since the things the change depends on aren't ready yet 16:25:05 geppetto: so I expect to see Koji able to do that by the time Fedora 34 lands 16:25:06 The big problem I see is what people do instead … getting rpm changes in the distro. isn't trivial. The lua mini package DB hasn't moved in 3 months? 16:25:35 Might be safe to wait until the oldest stable release is f34, so people keeping branches in sync don't shoot themselves in the foot. 16:25:43 safer^ 16:26:15 I suppose that if koji switch, it will affect all buildroots, no? 16:26:21 yes 16:26:28 koji changes affect all build roots 16:26:50 basically we need to have this fixed in koji before builders are reprovisioned with Fedora 33 or 34 16:26:57 We don't usually wait until something is available in every supported release. 16:27:00 definitely before F34, as we can hack around it for F33 16:27:18 tibbs: Yeh 16:27:19 But it's good to wait until things are at least close to being available in one supported release. 16:28:01 I mean I think it's likely that the buildroot changes to make rpm calls happy will happen before any other alternative gets implemented 16:29:28 I think the best course of action is to make sure the bootsrap mode will actually be enabled in koji 16:29:54 King_InuYasha: Can you point to any koji issue where this is being tracked? 16:30:01 let me see if I can find it 16:30:02 one sec 16:30:09 👍 16:31:01 https://pagure.io/koji/issue/1017 16:31:14 hey look, it's fixed already in koji: https://pagure.io/koji/c/1f1da58 16:32:01 hah, I should know this, since I reviewed it :P 16:32:21 that shows me and my scrambled brain :P 16:32:36 * limburgher hands King_InuYasha a coffee 16:32:47 Is that deployed? 16:32:52 yes 16:33:00 supported in koji 1.22 which was rolled out a couple weeks ago 16:33:01 https://docs.pagure.org/koji/release_notes/release_notes_1.22/#builder-changes 16:33:07 but the setting is not turned on yet? 16:33:18 pretty much 16:33:22 it's not configured, so koji doesn't use it 16:33:53 so koji and copr are in the same situation at this point 16:33:54 #info This is already fixed in upstream koji, but the fix is not turned on yet in fedora infra (must be on by F34) 16:33:57 just need to be configured 16:34:35 King_InuYasha: Can someone volunteer to poke both services to change sooner rather than later? 16:34:47 I can talk to copr folks 16:35:00 I talk to them semi-frequently already for other reasons 16:35:19 #action King_InuYasha will volunteer to talk to copr to get config. turned on 16:35:25 Anyone for fedora infra? 16:36:50 Sorry, I'm reading the docs. 16:37:27 carlwgeorge: do you talk to fedora infra guys? 16:37:41 So koji needs mock.use_bootstrap=1 set on the relevant tags. 16:37:50 tibbs: yes 16:37:50 I talk to them often enough. 16:37:51 I imagine Fedora infra (CPE) will punt on that since koji is categorized as an app we run but don't maintain 16:38:00 O.o 16:38:10 who, umm, maintains it? 16:38:12 This whole CPE thing is... something. 16:38:27 I would say that's a releng thing, but it's OK, I talk to them as well. 16:38:38 The remaining question is what the relevant tags would be. 16:38:49 #action tibbs will volunteer to talk to releng to get config. turned on 16:38:58 tibbs: I'd assumed F34+ at least 16:39:13 actually, probably the other direction once builders are F33 16:39:14 at least that should cause the least friction 16:39:30 since host would be sqlitedb and chroots would be bdb 16:39:34 Ahh 16:39:47 the straightforward answer is to just do it for all of them 16:39:59 mboddu might know the best person to make that call for fedora's koji instance 16:40:04 yeah 16:40:15 * mboddu reading back 16:40:38 maybe somebody should make this a FNaN System Wide Change? 16:41:16 decathorpe: it's a builder change though, not a distro. change 16:41:32 Although I guess the mock default should change for F34 too 16:41:39 mock *already* does bootstrap by defualt 16:41:43 *default* 16:41:44 ha 16:41:50 * decathorpe shrugs, hasn't stopped us before 16:41:52 fair enough 16:42:03 mock started doing bootstrap by default in version 2.0, iirc 16:42:19 decathorpe: eh, the FNaN process mostly confuses me 16:42:28 https://github.com/rpm-software-management/mock/wiki/Release-Notes-2.0 16:42:49 * geppetto is kind of confused why everyone else didn't pick it up at that point 16:42:54 Oh, well. 16:43:03 that's why I thought copr already did this 16:43:09 because it was the copr devs that changed it in mock 16:43:49 geppetto: koji hand-holds mock, so mock defaults don't necessarily apply to koji in all cases 16:43:55 weird, but whatever 16:44:26 Makes some sense for stability. 16:44:52 glancing at build logs makes it clear that koji leaves very little up to mock defaults (lots of flags are passed) 16:45:02 But for this, I'm not sure what the side effects might be or whether any of them might be negative. 16:46:08 tibbs: I think most bug side effects have been solved since it's been the default for local builds for ages 16:46:39 yup 16:46:40 (not accounting for weird modes koji might run mock in) 16:47:01 the lack of a koji staging env has really made testing that stuff painful 16:49:27 Hmm, ok 16:49:39 tibbs or someone: Can you create a ticket about this at pagure.io/releng so that I can take a look at it and check it with Kevin and other folks before we make the necessary changes? 16:50:05 Probably we will set it in stg once thats available 16:50:56 I think we are mostly ok with this then now … do we want to vote to undo the previous vote? 16:51:53 Seems ok to me. 16:51:55 geppetto: yep 16:52:19 there is some ticket thou... let em check 16:53:36 Need to see Panu's statement that this is OK if the chroot is set up so that the databases are guaranteed to match. I wouldn't be for it otherwise. 16:53:55 Panu has been the one pushing for it the most 16:54:16 He was quite against it previously and that didn't have much to do with database stuff. 16:54:58 I see it now. 16:55:02 ha 16:55:02 https://pagure.io/releng/issue/9308 16:55:06 "If mock bootstrap-mode is used, the reason for avoiding rpm queries from the spec simply vanishes." 16:55:28 https://pagure.io/releng/issue/9308#comment-668692 16:56:13 "We might not technically need bootstrap to keep things limping along for the time being, but we want to enable it the sooner the better." 16:56:48 question from a noob: can I query bconds from rpm? 16:56:59 sort of? 16:57:22 I think the answer depends on how you interpret the question. 16:57:38 we are ta this topic for almost 1 hour 16:57:45 sorry… 16:57:57 mhroncok: True, but it looks like it's resolved now 16:58:03 it is 16:58:09 I think we know what needs to be done. 16:58:12 yep 16:58:15 tibbs: I want to find out if another package was built with tests or without tests 16:58:37 defolos: from the resulting RPM packages? I don't think so 16:58:51 defolos: uh I think the only way to do that would be to include RPM macros in built packages 16:59:00 unfortunately, the only environment flags recorded in the rpm headers right now are compiler flags 16:59:13 we don't (yet) record macros set in the build environment into the headers 16:59:20 hm, darn 16:59:24 #info Given the changes in koji/copr the need to ban rpm calls from build time specfiles is unnecessary 16:59:30 so back to integrating fm-orchestrator into koji 16:59:33 #topic Open Floor 16:59:45 So, we have 20 seconds left :) 16:59:58 please vote on the Python prs 17:00:04 in the tickets or here 17:00:10 defolos: echo "%_with_tests" > %buildroot/usr/lib/rpm/macros/macros.%name.tests ;) 17:00:33 https://pagure.io/packaging-committee/pull-request/1011 17:00:37 https://pagure.io/packaging-committee/pull-request/1013 17:00:38 ? 17:00:49 indeed 17:00:53 geppetto++ 17:00:53 mhroncok: Karma for james changed to 5 (for the current release cycle): https://badges.fedoraproject.org/tags/cookie/any 17:01:03 geppetto: there's some interest from the openSUSE side about collaborating on a shared packaging guidelines doc that can spit out Fedora or openSUSE guidelines by "building" with an argument 17:01:05 +1 17:01:16 since the differences between the two are increasingly minor 17:02:00 * defolos is in favor of that ;-) 17:02:02 I think we might actually want to get all of the conversion bugs out of our existing documents before we try to do that. 17:02:05 King_InuYasha: I think there might be some interest in adding macros or something so that the same specfile works in both … but a pre-processor step I can't see being accepeted 17:02:08 * jwf waves 17:02:17 Plus the bureaucracy would be kind of crazy. 17:02:20 jwf: Sorry, blame koji ;) 17:02:24 (should we move the fedora council meeting to #fedora-meeting?) 17:02:35 jwf: waves to say hello or to have another meeting? :D 17:02:39 mattdm: We'll be another minute … not longer 17:02:50 mhroncok: Another meeting 17:02:52 geppetto: I think it'd mostly be some extra pages for openSUSE specific things like group tags and such 17:02:53 geppetto: :) 17:02:55 Yes, let's stop. Can review those two PRs and comment in the ticket. 17:02:59 mhroncok: Both! 17:03:00 yeah I think we're done :) 17:03:05 anyway, we're done here 17:03:13 jwf: o/ 17:03:29 ok :) 17:03:30 * dgilmore glares at bcotton 17:03:31 Ok, everyone go vote on 1011 and 1013 now … 17:03:34 lol 17:03:41 geppetto: basically, we're at the point where openSUSE wants to refork fedora guidelines again... 17:03:43 #endmeeting