16:00:02 #startmeeting fpc 16:00:02 Meeting started Thu Apr 26 16:00:02 2018 UTC. The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00:02 The meeting name has been set to 'fpc' 16:00:02 #meetingname fpc 16:00:02 #topic Roll Call 16:00:02 The meeting name has been set to 'fpc' 16:00:06 Howdy. 16:00:09 'sup 16:00:09 * limburgher here 16:00:11 #chair tibbs 16:00:11 Current chairs: geppetto tibbs 16:00:13 #chair limburgher 16:00:13 Current chairs: geppetto limburgher tibbs 16:00:14 How do 16:00:17 #chair redi 16:00:17 Current chairs: geppetto limburgher redi tibbs 16:00:21 #chair mbooth 16:00:21 Current chairs: geppetto limburgher mbooth redi tibbs 16:00:32 hey peeps 16:02:35 #chair decathorpe 16:02:35 Current chairs: decathorpe geppetto limburgher mbooth redi tibbs 16:02:36 hi! sorry for being late 16:02:41 * Pharaoh_Atem waves 16:02:41 no problem 16:03:06 I give everyone until at least 5 past anyway :) 16:03:11 * decathorpe waves back 16:03:14 oh good :) 16:05:15 Hmm, I assume igor would be here … oh well. 16:05:37 #topic Schedule 16:05:40 https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/UATLIX2DFWOOQY2SDEAWENND3W6UGSLE/ 16:05:54 #topic #761 Modernize R guidelines 16:05:58 .fpc 761 16:06:02 geppetto: Issue #761: Modernize R guidelines - packaging-committee - Pagure - https://pagure.io/packaging-committee/issue/761 16:07:02 Ahh, we have talked about this before … so it's a followup … but still 16:07:16 did anybody find out why including a "%build" section is required? 16:07:17 Most of this is done; I just had the question about the empty %build section. 16:07:40 It's not required for anything as far as I can tell. I do not know why that requirement was in there originally. 16:07:56 My only guess is that it was related to rpmlint complaining about it. 16:08:00 in my experience it triggers -debuginfo subpackage generation 16:08:14 And as I wrote, I don't believe that's the case. 16:08:15 At all. 16:08:32 debuginfo subpackage generation is not hung off of anything related to %build. 16:08:50 did you check? 16:09:13 I encountered this exact behaviour not long ago. 16:09:16 I did some checks. I don't know what specific case you encountered so of course I couldn't check that to understand what might have happened. 16:10:03 It's certainly possible I'm wrong, but I have been through the RPM source related to this as well as all of the macros and I don't see how it would matter. 16:10:13 But in any case, if I understand correctly, your case was the opposite. 16:10:25 You had to remove build in order to get rid of an unwanted debuginfo package. 16:10:36 exactly 16:10:44 And we're talking about why %build might be required. 16:10:59 Note that arch-specific R packages just do the building in %install. 16:11:22 Which is how anything could get built when %build is empty. 16:11:46 and do -debuginfo subpackages get generated without %build present? 16:12:03 My understanding is that they do. 16:12:31 Anyway, it's not a big deal. I figure if any of us would have the inside scoop, it would be ignatenkobrain. 16:12:55 yeh 16:13:07 iirc, old versions of rpm hated having missing sections 16:13:25 I think here old means RHEL4 or so. 16:13:36 oooooooooold 16:13:40 yeah, we're talking < 4.4, I think? 16:13:42 I do not believe EL5 cares. 16:13:51 rpmlint still complains too :/ 16:13:59 Yeah, I don't know why we don't turn that off. 16:14:11 I can see how things might trigger on stuff being in _builddir … but I doubt they'd care if that happened in %build or %install … ofc. I wouldn't be surprised at anything to do with rpm building :-o 16:14:52 ok, so I'd approve removing that specific wording around an empty %build having to be present, pending verification that everything works as expected 16:14:53 So we have any actions or info for this ticket? 16:14:57 Note that there is a lot of env var setting that happens in %build (via %___build_pre, I think) but that is included again in %install. 16:15:12 geppetto: Nah, do more testing and report back. 16:15:18 the main difference at this point with %build and %install is creation of %buildroot automatically 16:15:28 tibbs: that would be an action ;) 16:15:58 It's a very interesting question and we should get an answer independent of what happens with the R stuff. 16:16:07 Pharaoh_Atem: yeh, and that might be the trigger that decathorpe saw 16:16:33 and that latter bit is actually in redhat-rpm-config ;) 16:16:52 yeah, clarification in general would be nice 16:17:04 But __spec_install_pre explicitly creates the buildroot. 16:17:30 And it includes all of %___build_pre as well. 16:17:32 Hmm, ok, fair enough. 16:17:40 ¯\_(ツ)_/¯ 16:17:53 Note that I'm looking at rawhide here, so something could have changed at some point. 16:18:21 Still, I'll do more investigation. Just have to build 20 or so archful R packages with and without %build and rpmdiff the results. 16:18:37 Anyway, we should probably move on. 16:18:39 sounds good! 16:18:58 #action tibbs vollunteers to build a bunch of R packages with and without %build to see if there's a difference. 16:19:12 #topic #754 Should py3-foo obsolete py2-foo (when py2-foo is removed)? 16:19:18 .fpc 754 16:19:20 geppetto: Issue #754: Should python3-foo obsolete python2-foo (when python2-foo is removed)? - packaging-committee - Pagure - https://pagure.io/packaging-committee/issue/754 16:19:24 oh boy 16:19:28 this one is a bit of a mess 16:19:45 it is 16:19:48 * decathorpe frowns 16:20:00 Ahh, I thought there was an update … but just me being late with the summary 16:20:37 No proposals in the ticket yet … so we can move on. 16:21:05 #topic #743 Add link to C/C++ build flag docs. in redhat-rpm-config 16:21:07 .fpc 743 16:21:09 geppetto: Issue #743: Add link to C/C++ build flag documentation in redhat-rpm-config - packaging-committee - Pagure - https://pagure.io/packaging-committee/issue/743 16:21:47 Again, I think this was mbooth not mboddu 16:21:53 Yeah. 16:22:04 just mb doing the wrong thing 16:22:44 Haha, I think the people are used to tab completion right after two letters :) 16:22:57 So, I guess no real update here. 16:23:00 * Pharaoh_Atem shrugs 16:23:03 it's a nice document? 16:23:11 mboddu: Yeh, you even get hit for 3 which should be banned ;) 16:23:44 geppetto: Lol :) 16:23:45 if he was mohanboddu it'd work out better 16:23:48 but that's soooo long 16:23:55 mboddu: Sorry, I'm obviously too far down the alphabet 16:24:07 Anyway, I will let you guys carry on :) 16:25:11 #topic Open Floor 16:25:29 Ok, so that's it for tickets with updates … anyone else have anything else? 16:25:35 Pharaoh_Atem, mbooth : :) 16:25:55 I'm about to start on the action items assigned to me two weeks ago 16:26:04 Ruby? 16:26:07 what was the action for the updated ruby guidelines? 16:26:22 i tested them with 2-3 packages and found an issue 16:26:46 (which should be fixed with an update to rubygems-devel) 16:26:46 I think ignatenkobrain did what he indicated he would do, and that ticket should have moved back to the meeting status. 16:26:55 Ahh 16:27:02 That was 710. 16:27:26 But I don't know what to make of the last couple days of discussion in that ticket. 16:28:21 O.o 16:28:53 #topic #710 Ruby packaging guidelines update 16:28:59 .fpc 710 16:29:01 geppetto: Issue #710: Ruby packaging guidelines update - packaging-committee - Pagure - https://pagure.io/packaging-committee/issue/710 16:29:22 Basically, for the issue that decathorpe found, was any change to the draft needed? 16:29:46 I don't have a problem with something breaking a small number of packages if the ruby maintainers are OK with it. 16:29:59 just a change to macros.rubygems shipped with rubygems-devel is necessary 16:30:10 and it affects only 2 packages according to vondruch 16:30:48 yeh, no update from churchyard … but I'm fine with it if decathorpe says it's good 16:31:15 well, the changes look sane, and I created 2 packages with the new guidelines 16:31:25 which worked well 16:32:20 tibbs: did ignatenkobrain post a summary in email or talk on IRC? 16:32:41 But again, I'm happy to +1 based on decathorpe's review. 16:33:09 I didn't see anything from ignatenkobrain 16:34:06 Hmm, ok. Anyone else want to vote now? 16:35:29 +1 from me, assuming the macros get updated 16:36:46 argh 16:36:47 I again forgot 16:36:47 .hello2 16:36:48 ignatenkobrain: ignatenkobrain 'Igor Gnatenko' 16:36:56 those guidelines were looking good 16:37:02 apart from what decathorpe found 16:37:05 about macro in parent dir 16:37:24 #chair ignatenkobrain 16:37:24 Current chairs: decathorpe geppetto ignatenkobrain limburgher mbooth redi tibbs 16:38:16 something is wrong with my IRC client 16:40:57 ignatenkobrain: last thing I got was "about macro in parent dir" 16:41:31 hmm, okay 16:41:32 so it's +1 from me 16:41:34 just to be clear 16:42:34 Ok, that's 3 … tibbs mbooth redi … any of you want to vote? 16:43:03 Sorry, someone in my office. 16:43:10 Give me a couple to catch up. 16:43:31 not sure I understand the issues completely. not against the changes though 16:44:15 I know absolutely nothing about ruby, let alone ruby packaging 16:44:44 summary of the issue I found: the location of the .gemspec file changed between old and new guidelines, and two macros in rubygems-devel have to be updated as well. 16:45:11 *to account for that change 16:45:31 sounds good. +1 with that change, based on your review 16:45:53 but I'm still mostly ignorant :) 16:46:54 +1 this has sat around for too long anyway. 16:47:03 Sure, +1 16:47:09 If we have to tweak then we can do that next week. 16:48:19 #action VOTE: Ruby packaging guidelines update (+1:6, 0:0, -1:0) 16:48:24 ok 16:48:30 #topic Open Floor 16:49:12 Ok, only 10 minutes left … so anything quick? 16:49:22 ignatenkobrain: Anything you wanted to talk about this week? 16:49:56 ignatenkobrain: any info why "%build" section could be magical? 16:50:12 I think I might have figured it out. 16:50:26 decathorpe: IIRC RPM adds debuginfo in there 16:50:27 sec 16:50:27 will find it 16:50:45 debuginfo generation is keyed off of whether %buildsubdir is defined. 16:50:54 And that happens internally to RPM, not in macros. 16:51:11 https://src.fedoraproject.org/rpms/redhat-rpm-config/blob/master/f/macros#_161 16:51:36 so it seems it's defined in %install 16:51:37 debuginfo construction occurs in the install phase 16:51:38 not in %build 16:51:50 Yes, it's all defined and inserted into the spec at %install time. 16:51:58 But it's keyed off of whether %buildsubdir is defined. 16:52:34 And that happens in build/parsePrep.c as far as I can tell, which confuses me even more. 16:52:55 but since the R packages do everything during %install, it will work fine 16:53:07 And... it doesn't. 16:53:38 decathorpe was entirely correct; if the empty %build is not present, %buildsubdir never gets defined internally by RPM and so debuginfo generation never happens. 16:53:46 oh 16:54:02 See how %install is defined: 16:54:07 I was right? haha 16:54:08 install %{?_enable_debug_packages:%{?buildsubdir:%{debug_package}}} 16:54:26 aha 16:54:40 well at least now you can add that to the guidelines as justifcation for requiring %build 16:54:58 Yes, it makes sense. 16:55:00 yes, good idea to document that somewhere. 16:55:02 this is messed up 16:55:14 the mechanism to activate it is keyed on %install 16:55:21 #info tibbs confirms %build needed for debuginfo generation 16:55:34 but the actual build is required by %build :/ 16:55:40 Well, the injection of it all happens in %install. 16:56:07 I just don't understand the mechanism behind how %buildsubdir gets defined. 16:56:32 I mean, I guess it's this in build/parsePrep.c: rpmPushMacro(spec->macros, "buildsubdir", NULL, spec->buildSubdir, RMIL_SPEC); 16:57:16 But but that's in the doSetupMacro function, so I still don't understand what %build has to do with it at all. 16:57:47 let me ask different question, does removal of that section break anything? 16:58:06 You mean the empty %build section? 16:58:18 If so, yes, you don't get debuginfo packages if it's not there. 16:58:38 To recap: archful R packages do all of their building in %install and have an empty %build section. 16:58:53 If you remove the empty %build section, no debuginfo generation happens. 16:59:13 So the first question was whether the empty %build is required. That's a big yes. 16:59:27 The followup question is "WTF?" 16:59:35 haha yeah 16:59:49 info: we have anothere meeting here in 1 minute :-) 16:59:50 Anyway, we're at an hour and I learned something today. 16:59:55 at least we know now that the empty %build is there on purpose 17:00:06 I'll summarize in the ticket and close it out. 17:00:16 +1 17:01:59 * nirik looks for the F28 go/no-go meeting... if FPC is done? 17:02:01 * jkurik propose #endmeeting :-) 17:02:13 ok 17:02:18 #endmeeting