17:00:11 #startmeeting fpc 17:00:11 Meeting started Thu Nov 29 17:00:11 2018 UTC. 17:00:11 This meeting is logged and archived in a public location. 17:00:11 The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:11 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:00:11 The meeting name has been set to 'fpc' 17:00:12 #meetingname fpc 17:00:12 The meeting name has been set to 'fpc' 17:00:12 #topic Roll Call 17:00:18 Hey folks. 17:00:27 #chair tibbs 17:00:27 Current chairs: geppetto tibbs 17:00:30 * limburgher here 17:00:38 .hello2 17:00:41 ignatenkobrain: ignatenkobrain 'Igor Gnatenko' 17:01:10 #chair limburgher 17:01:10 Current chairs: geppetto limburgher tibbs 17:01:13 #chair ignatenkobrain 17:01:13 Current chairs: geppetto ignatenkobrain limburgher tibbs 17:01:45 * decathorpe was stuck in traffic, will be at PC in a few minutes 17:01:57 ok 17:02:00 #chair decathorpe 17:02:00 Current chairs: decathorpe geppetto ignatenkobrain limburgher tibbs 17:02:06 * handsome_pirate shows up 17:02:52 apologies, but I have childcare duties today, so can't attend this meeting 17:03:05 I'll read through the logs later though 17:04:06 ok, no problem 17:06:01 decathorpe: let us know when you get there 17:08:27 * decathorpe is here 17:09:09 ok 17:09:13 #topic Schedule 17:09:14 https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/ZDZVTB3TUYZQGUTEGMA5377OR54VOCO2/ 17:09:25 #topic #806 Guideline change procedure should be updated for PRs 17:09:30 .fpc 806 17:09:32 geppetto: Issue #806: Guideline Change Procedure should be updated for Pull Requests - packaging-committee - Pagure.io - https://pagure.io/packaging-committee/issue/806 17:10:26 I feel like we've talked about this, but I don't see anything 17:10:33 I guess we decided last week that we need to have the issue opened if there's going to be anything to vote about. 17:11:09 And the linked guidelines change procedure section does say to open an issue. 17:11:12 I'm in favor in principle, but I'd love to see text submitted for bikeshedding. 17:11:38 Well I made the changes weeks ago. 17:11:45 https://fedoraproject.org/wiki/Packaging_Committee#Guideline_Change_Procedure 17:12:18 Well looky there. 17:12:23 #info Decided on a week or two ago, need to open issues if there's anything to vote on. 17:12:37 #topic #830 Update Python manual byte compilation rules 17:12:42 .fpc 830 17:12:43 geppetto: Issue #830: Update Python manual byte compilation rules - packaging-committee - Pagure.io - https://pagure.io/packaging-committee/issue/830 17:13:00 I was +1 in the ticket. 17:13:25 I think we've already discussed about this some months ago? 17:13:31 I'm +1 17:13:32 I'm =1 17:13:37 Yeah, we discussed the first half of it. 17:13:37 Or actually +1 17:13:54 yeah, I think we decided to finally flip the switch for f30 later (later seems to be now) 17:13:59 so I'm +1 17:14:11 Well, it wasn't us that decided to flip the switch. This was a feature. 17:14:17 The guidelines change is just reflecting that. 17:14:21 right 17:14:22 +1 17:16:55 #action Update Python manual byte compilation rules (+1:6, 0:0, -1:0) 17:17:18 #topic #834 Multilib file conflict 17:17:20 .fpc 834 17:17:23 geppetto: Issue #834: Multilib file conflict - packaging-committee - Pagure.io - https://pagure.io/packaging-committee/issue/834 17:18:12 That's a bizarre ticket. 17:18:14 Can we actually do anything non-stupid about this? 17:18:41 Says there's an issue with the guidelines, but then it's really a bug in something at the dnf level. 17:19:01 I'm not sure dnf actually does anything different to yum? 17:19:04 And from what I gather, even what they propose won't actually fix the problem. 17:20:11 I wonder how many packagers would opt out of multilib if they had the option. 17:20:15 ignatenkobrain: What does dnf do differently? And is there anything we can do? 17:20:33 yum always made sure that multilib packages are in sync 17:20:38 dnf doesn't 17:20:40 tibbs: I don't think they can opt out … but at this point I'd guess all of them 17:20:48 ignatenkobrain: ahh 17:21:04 I'm not sure if there used to be an option to tweak this behavior in yum, but the default was to keep them in sync 17:21:09 ignatenkobrain: is that not somewhat easy to work around on the dnf side? 17:21:15 Certainly not all of them. There are a few things where it's still useful to have 32bit libs around. But they're getting rarer these days. 17:21:36 libsolv ticket has some information how to workaround it before there will be an option in libsolv 17:21:50 ah, so the issue is: "foo.x86_64" doesnt conflict with "foo.i686" != version-release? 17:21:53 but dnf guys never tried to play with it 17:21:56 decathorpe: yes 17:21:58 ignatenkobrain: like adding an implicit "Conflicts: %{name} != %{version} 17:22:07 geppetto: no 17:22:17 but similar 17:22:44 Simple conflicts will make it worse because it works on provides too 17:22:55 #info Main problem is that DNF is much happier than yum to let foo.x86_64 and foo.i686 be different versions 17:23:22 Error: Protected multilib versions: audit-libs-2.4.1-1.fc20.i686 != audit-libs-2.4-2.fc20.x86_64 17:23:23 #info Working around this in packages is generally agreed to be insane 17:23:28 this was the message back in yum times 17:23:47 * geppetto nods 17:24:08 I remember writting that, for this very reason … and it could be turned off by an option. 17:24:51 so the "workaround" exists, but no one ever tried to play with it in libdnf to my knowledge 17:24:52 ignatenkobrain: So … is there anything FPC can do to help here? 17:25:08 I think only to recommend DNF guys to actually try the option mls proposed 17:25:25 and myself I will make sure that such option lands in libsolv upstream 17:25:26 Does RPM actually have a != comparison operator? 17:25:36 tibbs: yeh 17:25:50 tibbs: yes, `(foo > 1-1 with foo < 1-2)` ;) 17:25:56 I don't think we can help in any way that's actually helpful. 17:26:00 or the other way around 17:26:35 I wonder if a package can even conflict with other versions of itself. 17:26:50 It still seems like the kind of thing a packager really shouldn't have to worry about. 17:27:14 agree 17:27:23 tibbs: 100% … the conflict != thing should be done automaticcaly in some layer of dnf 17:27:43 Or at least at some layer of RPM that the packager doesn't have to care about. 17:27:57 sure 17:28:21 geppetto: can you #info option in yum please? 17:28:51 ignatenkobrain: protected_multilib in "man yum.conf" 17:29:11 #info The yum option is "protected_multilib". 17:29:56 yeah, such option is not implemented in dnf as I can see 17:30:22 #action There's nothing useful FPC can do here, there are multiple options to fix the dnf/rpm stack and we encourage the developers to implement one 17:30:37 Ok, now the big one: 17:30:39 #topic #832 Proposal for version guideline overhaul (including tilde) 17:30:42 .fpc 832 17:30:43 geppetto: Issue #832: Proposal for versioning guideline overhaul (including tilde) - packaging-committee - Pagure.io - https://pagure.io/packaging-committee/issue/832 17:30:53 tibbs: You want to say anything? 17:31:23 I think I've said too much already. 17:31:47 Support for '^' in RPM version comparisons was merged upstream. 17:32:03 It may make it into Fedora in... a few months. 17:32:09 it's looking good, and I'm looking forward to ^ support when it can be cleaned up further 17:32:59 I like tibbs latest proposal with tilde 17:33:02 In the meantime, what I figured we could do is come up with a "least change" proposal which allows tildes in specific circumstances. 17:33:02 it is much cleaner and easier 17:33:33 And then when we get caret in general Fedora releases, come back and look at adopting it. 17:34:02 I really think it makes things nice and clean, but right now it's not even worth discussing because even rawhide doesn't have it. 17:34:21 And RPM devs object to any backport of it, so that won't be happening. 17:35:10 I'd rather wait until we could clean it up fully 17:35:26 so we should document ^ as an issue for future FPC 17:35:29 but if everyone else wants to do it without ^ I'll go along 17:36:52 geppetto: I agree with that sentiment, which is why I made the proposal I made. 17:37:24 * geppetto nods 17:37:53 Just a minimal patch to the guidelines which allows tilde as long as you don't mix it with snapshots. That's basically what the people who are using it appear to want. 17:38:34 And note that depending on some other "fun" decisions, we could potentially need to wait until RHEL8 goes EOL before we could adopt caret. 17:39:16 sigh 17:39:28 can we document that somewhere, so it won't be forgotten in 20 years? 17:39:49 Having different versioning guidelines between Fedora and EPEL is something we'd have to think about if it came down to that. 17:40:08 That really seems like a change that rpm devs should just get in now to not cause everyone pain for the next decade. 17:40:17 But given that we don't even have the thing in rawhide, it's a bit early to waste time discussing. 17:40:42 Well, RPM is RPM. And RHEL is RHEL. I certainly don't have any influence over either. 17:40:45 yeh, not even in rawhide is just facepalm 17:41:08 Well, it just got merged upstream so we do have to give it a bit. 17:41:33 It is unfortunate that ignatenkobrain actually came up with this two years ago but there was no perceived need for it so it never got merged. 17:42:41 So anyway, what I'm basically putting on the table right now is https://fedoraproject.org/wiki/User:Tibbs/TildeVersioning2 17:43:17 And the diff for that is https://fedoraproject.org/w/index.php?title=User%3ATibbs%2FTildeVersioning2&type=revision&diff=529479&oldid=528965 17:43:32 Yes, this is all still in the wiki because that's where all of this work lived originally. 17:44:22 > * the version that the next actual release will take ( 17:44:26 need closing brace 17:44:43 This draft covers both the requested tilde cases as I've understood them, and the use of tilde within %dist that the bootstrapping macros use. (I think.) 17:44:49 > phgrel 17:44:53 there is a typo ;) 17:45:47 tibbs: by which case do we allow tilde in Release for ~bootstrap? 17:46:22 "Note that the Dist tag is supplied by other portions of the system and may in some circumstances contain additional structure, including tildes. As this is not under the control of the packager, that structure is not covered here. The packager MUST simply include %{?dist} verbatim as indicated above. " 17:46:34 I'm +1 on the diff 17:46:48 +1 too 17:47:34 me+1too 17:48:01 I'm +1. 17:48:14 tibbs: are you going to convert it to asciidoc? 17:48:18 Yes, of course. 17:48:19 or should I do that for you? 17:48:56 No, I can do it. It's not hard to just patch this over the current versioning guidelines (which just got cleaned up the other day). 17:49:33 limburgher: vote? 17:49:35 I did fix the typos. And then made a typo in the commit message. Totally on purpose. 17:49:45 :) 17:49:49 :D 17:50:09 The one thing I fear is that someone will complain that this doesn't cover their use case. 17:50:26 then let them complain. this is a good solution for now 17:50:54 There are other ways I could take this further, of course, but I really wanted to make the fewest actual changes. 17:51:23 I did write something on the order of a dozen drafts to get to this point. 17:52:01 this should cover main use-cases for tilde 17:52:10 ~bootstrap and 1~rc1 17:52:20 I believve that's enough for start 17:53:23 Looks like limburgher might have fallen off chat, we can leave it for more votes in the ticket or another week … I think churchyard will probably +1 anyway 17:53:32 Yes, and then we have time to see what's going to happen with caret and then decide if we want to do the big overhaul. 17:53:39 #info Tibbs minimal proposal for tilde in versioning (+1:4, 0:0, -1:0) 17:53:48 #action More voting in the ticket, or another week 17:53:51 #topic Open Floor 17:54:05 Ok, so I'm very likely not going to be here next week 17:54:15 ignatenkobrain: Can you run the meeting? 17:54:25 If not then I can ask churchyard 17:54:47 geppetto: let me check calendar 17:54:54 I have to take my wife to the airport on the 6th. 17:55:12 geppetto: can't promise yet, because we have table footbal tournament ;) 17:55:19 cool :) 17:55:23 I have to get her there just as the meeting would be starting. 17:55:36 geppetto: I will let you know on monday 17:55:39 Probably won't be 5 people anyway, so we can just skip it. 17:55:42 and if I can't, I will ask Miro 17:55:42 So I would most likely be late unless she decides to get her there really early. 17:55:57 * geppetto nods … don't worry about it 17:56:15 we dealt with all the new tickets this week, and it's often slow around the holidays 17:56:21 if it's questionable that most of us will be there on time or at all, we could just skip next week 17:56:31 * geppetto nods 17:57:32 true 17:58:03 (it would also give me more time to finally look at forge macros) 17:58:11 It's better to plan skipping than skip planning. 17:58:11 Ok, unless there'a anything else I'll end the meeting exactly on time :) 17:58:17 Hooray. 17:59:24 decathorpe++ 17:59:24 nim: Karma for decathorpe changed to 1 (for the current release cycle): https://badges.fedoraproject.org/tags/cookie/any 17:59:24 #endmeeting