18:01:32 #startmeeting fpc 18:01:32 Meeting started Wed Feb 7 18:01:32 2018 UTC. The chair is geppetto. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:32 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:01:32 The meeting name has been set to 'fpc' 18:01:32 #meetingname fpc 18:01:32 #topic Roll Call 18:01:32 The meeting name has been set to 'fpc' 18:01:39 hello ;) 18:01:47 #chair tibbs 18:01:47 Current chairs: geppetto tibbs 18:01:49 #chair Rathann 18:01:49 Current chairs: Rathann geppetto tibbs 18:01:55 #chair mbooth 18:01:55 Current chairs: Rathann geppetto mbooth tibbs 18:01:59 #chair orionp 18:01:59 Current chairs: Rathann geppetto mbooth orionp tibbs 18:02:16 Well, I'm obviously organized today … trying to have the meeting in $fedora-devel :) 18:04:14 no worries 18:04:17 * limburgher here-ish 18:04:17 I'm trying not to get the flu today. 18:04:33 * Rathann trying not to fall asleep 18:05:14 #chair limburgher 18:05:14 Current chairs: Rathann geppetto limburgher mbooth orionp tibbs 18:05:22 6 :-o 18:05:37 #topic Schedule 18:05:38 https://lists.fedoraproject.org/archives/list/packaging@lists.fedoraproject.org/message/NJL2IBBF6HN622BZ5AVI4E5TDZC7SZ5A/ 18:05:52 #topic #743 Add link to C/C++ build flag docs. in redhat-rpm-config 18:05:52 .fpc 743 18:05:56 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 18:06:24 this seems reasonable 18:06:55 I'm tempted to say "yes, please do a draft" :) 18:06:59 LOL 18:08:26 I suspect florian is super busy … I guess if we just add the link somewhere, we can probably hack it in "now" 18:08:39 I swear I had commented in that ticket. 18:09:07 in this section: https://fedoraproject.org/wiki/Packaging:Guidelines#Compiler_flags 18:09:30 Well.... one interesting question is whether we're comfortable with the document we're linking to basically being about the current state of rawhide. 18:10:02 I thought that'd gone everywhere now 18:10:05 And also, at what point does content in a linked document stray into guidelines territory. 18:10:41 ignatenkobrain: is the new redhat-rpm-config config. in all releases or just rawhide? 18:11:04 Well, the point is that it links straight into git. 18:11:05 I propose adding something like "Current (rawhide) compiler and linker flags are documented here: [link] or even turn "system rpm configuration" into a link. 18:11:14 It's not pointing at a file that's on end user systems. 18:11:22 * geppetto nods … yeh 18:11:36 Note that Florian wrote that document basically because modifying the packaging guidelines is too difficult. 18:11:40 just rawhide I suppose 18:11:55 Really that document _is_ a packaging guideline. 18:12:13 better to ask fweimer if he wants to backport it to older releases... but I guess he doesn't want 18:12:26 I do think there's a case to be made for just importing it, but then modifications have to go through bureaucracy. 18:12:27 yeh, damit ... I first thought it was just a macro file with comments 18:13:00 No, it's a really nice guideline drafts. 18:13:21 And it does say that "you must" do things. 18:13:32 yeh, scimming the text it seems fine … but we can't just point to that, sigh. 18:13:50 I think that's really the fundamental question. 18:15:33 Personally I think it's a well-written document that we could certainly just import wholesale. 18:15:39 hm you may be right 18:16:18 Any idea what to call it? 18:16:31 Like replacing the whole c/c++ page with this doc, and removing the compiler flags section from the main guidelines? 18:17:34 Probably makes sense to also rename the page to "Packaging:C, C++ and Fortran" in that case 18:17:39 There's probably a little bit more in the c/c++ page which we'd want to keep, but yeah. 18:18:57 ok, so we really need a draft then … to see where this is going and what it is replacing 18:19:19 anyone want to volunteer to help florian? 18:20:04 geppetto: Assign it to me and I'll hopefully have some time 18:20:49 #action mboddu Will help florian get a real draft we can vote on, probably merging that docs. into the C/C++ page. 18:21:06 #topic #749 Add note about disabling BRP scripts to guidelines 18:21:08 by the way, I'm wondering why our C and C++ guidelines seem to suggest clang is an equal alternative to gcc? 18:21:13 .fpc 749 18:21:19 geppetto: Damn overloaded username prefixes ;-) 18:21:22 geppetto: Issue #749: Add note about disabling BRP scripts to guidelines - packaging-committee - Pagure - https://pagure.io/packaging-committee/issue/749 18:21:36 Gah 18:21:55 Rathann: Because is the main guideline it says you can use clang if your package does not compile with gcc, for example 18:22:26 they do, but the C and C++ page doesn't seem to prefer gcc over clang 18:23:09 #749 was something I wrote following the last meeting -- instead of various pages explaining how to disable burp scripts, I thought it could be in one place and referenced 18:23:51 Is there a good way to get a list of all the current brp scripts? 18:24:21 geppetto: https://src.fedoraproject.org/rpms/redhat-rpm-config/blob/master/f/macros#_137 18:26:48 geppetto: rpm --showrc, look for __os_install_post 18:27:09 That was the subject of my comment in 749. 18:27:16 $ rpm --eval '%__os_install_post' 18:27:57 Those are the expanded macros though 18:29:00 RIght, --eval isn't what you want. --showrc is the best thing (besides just looking in the macro files). 18:30:37 even showrc is after the macro's are expanded 18:30:54 Not on F27 and rawhide, at least. 18:31:21 What does: rpm --showrc | fgrep brp say? 18:31:35 Or at least, not in the way that matters for the question at hand. 18:32:33 Well I want something we can show packagers which is "you need to undefine foo" and to get a list of foo run XYZ 18:32:40 right 18:32:55 If that's what we want then we will have to maintain that list by hand. 18:33:23 On rawhide, __os_install_post just tells you things like "%{?__brp_ldconfig}". 18:33:58 The list shown there is all that will be called as far as I'm aware (and of course some of those are also conditional on __debug_package. 18:34:18 this seems to be the right file /usr/lib/rpm/redhat/macros 18:34:26 Yeh, was was just wondering if that file mbooth linked to is available somewhere you can fgrep _brp_ on 18:34:44 But I don't think it is … looks like it's compiled into rpm magically 18:34:55 geppetto: Is this not enough? https://paste.fedoraproject.org/paste/08IeB8hOvosoZmpR11AFCg 18:35:22 or /usr/lib/rpm/platform/%{_arch}-linux/macros 18:35:28 Please note that what you get on rawhide is significantly different than what you get on F27. 18:35:47 Yeah, I mean the guideline I wrote says F28+ 18:36:09 mbooth: I get '' for that on F27 18:36:42 If that's the way to see it going forward I'm happy to point to that fgrep … just that it doesn't do anything here. 18:37:27 F27 doesn't have any uniform method of disabling brp scripts. 18:38:05 Which is presumably why the proposal says that it's F28+ only. 18:38:50 * geppetto nods … looking at rawhide now 18:40:43 Cool, rawhide repo. is broke on F27 18:41:18 upgrading to rawhide rpm didn't show anything 18:41:48 upgrading to reawhide redhat-rpm-config ftw 18:42:34 It's all in /ur/lib/rpm/redhat/macros. 18:42:35 since I need to leave, I probably want to ask you to fix https://fedoraproject.org/wiki/Packaging:Scriptlets#Shared_Libraries page... it should use `libs` instead of `-n libs`, `%license` instead of `license` and `[…]` instead of `[,,,]` 18:43:32 Did I get that that wrong? The lack of uniformity where you use -n and where you don't in RPM is really annoying. 18:44:20 But wait, you do use %post -n foo. 18:44:28 So why not %ldconfig_scriptlets -n foo? 18:44:49 That's the way I intended the macro to work. 18:44:49 So this is probably what we want to put in: fgrep '%__brp_' /usr/lib/rpm/redhat/macros 18:45:17 ignatenkobrain: I fixed the missing % in %license. 18:45:29 I don't see what else you wanted to be changed. 18:46:10 Adding the fgrep line in, I'm +1 on the change for brp stuff. 18:46:52 geppetto: There's no guarantee that every mention of __brp_ is going to show a script that actually gets called by default. 18:47:05 Looking at __os_install_post is really the only reasonable way. 18:47:23 tibbs: s/-n libs/libs/ 18:47:32 Rathann: I disagree with that, though. 18:47:52 uh why? 18:47:59 That's technically true … but I hope they'd keep the same naming convention, so it would work … and this way we can say "run this command" instead of "hunt through this source file" 18:47:59 %files -n libs is surely wrong? 18:48:00 -n libs works fine and mirrors all of the other scriptlets. 18:48:19 So which -n libs are you talking about, then? 18:48:32 the one next to %files 18:48:53 So the libs subpackage doesn't have any files? 18:49:13 I mean, if there's a libs package at all, it has to have a %files section. 18:49:27 exactly 18:49:38 And it will start with %files -n libs, won't it? 18:49:43 but %files -n libs is for a package name 'libs', not 'foo-libs' 18:50:09 Rathann: yep 18:50:21 So you're asking for libs to be changed to example-libs or whatever? 18:50:42 no, I'm asking for s/%files -n libs/%files libs/ 18:50:43 or just drop `-n` in there 18:51:35 and s/[,,,]/[...]/ also 18:51:36 So I guess 749 got derailed completely, sorry. 18:51:48 Rathann: You know you can edit that page.... 18:51:55 :-) 18:52:05 I thought the idea of "libs" there was meant to imply "the package which provides the libraries" not necessarily "foo-libs" 18:52:28 So I don't think it's wrong … but maybe it's better without the -n 18:52:34 * geppetto ¯\_(ツ)_/¯ 18:52:44 heh 18:53:10 There are certainly examples of subpackages which aren't just %{name}-libs 18:55:36 Sadly we're still not getting through enough stuff. 18:56:30 Having a meeting is a start, wanting to decide on what to do for more than one ticket is just optimism 18:57:02 Anyway I think people mostly agree with 749 18:57:11 Yes. 18:57:16 I voted already 18:57:42 So that's +3 18:57:54 I updated it to include your suggestions 18:57:56 I'd like a list command, but if you'd prefer not to … then I'm +1 anyway 18:58:05 So that's +4 18:58:06 I'm +1 to any of the three proposed variants (the original, mine that mentions os_install_post, and one that gives more detailed info on how to find out what will be called. 18:58:20 orionp: limburgher: Vote? 18:58:35 i.e. mention _os_install_post so packagers can go digging AND mention the fgrep command to list them on F28+ 18:58:44 Hopefully that pleases all :-) 18:59:34 sure, +1 18:59:40 Sorry, was AFK. . .reading. . . 18:59:57 +1 19:00:01 Ok, that's +5 19:00:24 #action Add note about disabling BRP scripts to guidelines (+1:5, 0:0, -1:0) 19:00:35 So … done, two tickets dealt with (kinda) 19:00:44 #topic Open Floor 19:01:05 Anything anybody wants to talk about in the -0 seconds we have left ?:) 19:01:47 nope 19:02:20 #endmeeting