22:02:09 #startmeeting EPEL (2020-12-18) 22:02:09 Meeting started Fri Dec 18 22:02:09 2020 UTC. 22:02:09 This meeting is logged and archived in a public location. 22:02:09 The chair is tdawson. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:02:09 Useful Commands: #action #agreed #halp #info #idea #link #topic. 22:02:09 The meeting name has been set to 'epel_(2020-12-18)' 22:02:11 #meetingname epel 22:02:11 The meeting name has been set to 'epel' 22:02:12 #chair nirik tdawson bstinson pgreco carlwgeorge michel_slm 22:02:12 Current chairs: bstinson carlwgeorge michel_slm nirik pgreco tdawson 22:02:14 #topic aloha 22:02:18 .hello salimma 22:02:19 michel_slm: salimma 'Michel Alexandre Salim' 22:02:26 Hi michel_slm 22:02:42 * cyberpear listens in again 22:02:43 hi tdawson ! 22:02:49 * King_InuYasha waves 22:02:50 Hi cyberpear 22:02:51 .hello ngompa 22:02:52 King_InuYasha: ngompa 'Neal Gompa' 22:02:55 Hi King_InuYasha 22:03:06 * carlwgeorge waves 22:03:12 Hi carlwgeorge 22:04:37 hello! 22:05:17 Hi pgreco 22:05:39 Quite alot for potentially the last meeting of a crazy year. 22:05:49 #topic Old Business 22:05:50 EPEL-Packaging-SIG 22:06:12 * King_InuYasha sighs 22:07:02 michel_slm: What do we have left to do on the EPEL-Packaging-SIG now that the documentation is done? 22:07:17 one sec, cat's attacking my keyboard 22:07:36 Gotta hate that. 22:07:43 he's so cute I can't get mad 22:08:04 * michel_slm 's cat likes to massage his back with keyboard keys 22:08:06 I'm sure the cat has less typos than hughesjr :) 22:08:22 Do you need me to do epel8-next first? 22:08:25 lol. I think we can link the Packaging page to the main EPEL wiki and announce it 22:08:30 nah, I chased the cat out 22:08:53 also we can take a look at the queue of requests and see what might make sense for the SIG to own 22:09:22 https://bugzilla.redhat.com/buglist.cgi?bug_status=NEW&f1=delta_ts&list_id=11543139&o1=greaterthan&order=Last%20Changed&product=Fedora&product=Fedora%20EPEL&query_format=advanced&short_desc=epel&short_desc_type=allwordssubstr&v1=-1w 22:10:10 (maybe I'll just post this to the epel-devel list on a periodic basis rather than doing it during the meeting) 22:10:57 one question for now though - if we want to be able to get the SIG added to a package without triggering the nonresponsive maintainer process, where should we file this? 22:11:05 fesco maybe? 22:12:34 Sorry, was looking at the list 22:12:36 that's usually fesco approval and then releng does the thing 22:14:09 ok, I'll get that started then. was hoping to get the ability for collaborators to branch a package first but that seems to be taking a long time 22:14:32 I just noticed that that list is for just the past week. I bet if we go back several weeks, we would be able to invoke the "unresponsive maintainer" and just take them. 22:14:43 in general people have been added as committers when they want to maintain for EPEL only, so most maintainers are probably ok with it 22:14:54 Yep 22:14:58 tdawson: yeah, I can change that to be > 2 weeks for now 22:15:21 eventually we want to do 1 week (at least for the packages we care about) since the goal is to speed up the process 22:15:35 Ya, I'm betting that most maintainers would add the sig if we asked. 22:15:43 as a backup 22:16:06 ok, I think that's it for me on this topic. I'll also use epel tickets for proposed documentation changes now that it's live 22:16:14 pgreco: ? 22:16:33 michel_slm: as a backup for them, in case they don't respond in time I meant 22:16:39 ah, yes 22:16:58 if the sig is added, we need to try to maintain the order for those who keep all their branches in sync 22:17:14 we need to be respectful of that 22:17:49 as in, not push changes to the Fedora branches without a PR first? 22:18:41 that, or as in not push changes that would make the branches diverge 22:18:58 like what happened with the packages.cfg 22:19:10 yeah. 22:19:22 which had the side effect that every time the maintainer merged to epel8, it had a merge commit 22:19:38 instead of a fast forward, like I like to keep in my branches 22:19:45 I'll propose a code of conduct for packager SIG members to reassure maintainers we won't do this 22:19:46 * nirik arrives late 22:19:59 Yep, that made epel8 unpopular among several. 22:20:02 Hi nirik 22:20:14 michel_slm: that we'll try to be mindful of it ;) 22:20:24 reassure is a strong word 22:21:08 yeah 22:21:09 Although I think most maintainers who do all branches at once, probrubly don't want/need the sig as a backup. They usually have it under control. 22:21:32 * pgreco nods 22:21:36 yup, it would be like the Python SIG - if you think you need help maintaining the package, add us 22:22:09 it's the existence of maintainers that ignore EPEL requests that this is meant to help with 22:22:54 OK, I'm going to move on. 22:23:14 epel8-next - carlwgeorge do you have a status for us? 22:23:23 no news on the epel next front, still waiting on staging bodhi permissions. i fear any further work is going to get pushed back till after shutdown. 22:23:34 carlwgeorge: And no, I didn't get an rpm made that has that feature. 22:24:10 i think King_InuYasha may be able to help here, i can get with him after the meeting. we can test that out while other stuff is blocked. 22:24:52 * King_InuYasha nods 22:25:00 Has anyone in this group seen the rpm Requires feature/option that basically is like a runtime if statement "if package X is installed, then I require package Y, otherwise don't worry about Y" 22:25:51 i.e. in epel-release, if centos-stream-release is installed, require epel-next-release 22:26:02 yup 22:26:20 I think I saw it in either redhat-rpm-config or python-rpm-macros 22:26:22 I saw that recently, was all excited ... and now I can't remember which package I saw it on ... so I have no example. 22:26:23 i've heard of such things but i'm not familiar with the implementation, or if el8's rpm is new enough 22:26:49 Hmmm ... I've looked at both of those recently ... I'll look there. 22:26:51 should be easy enough to do... although... 22:26:58 I seem to remember some (if gcc require annobin) or something like that in spec files 22:27:03 tdawson: yes, I've used it plenty 22:27:14 rpm 4.14 in el8 has it 22:27:15 https://src.fedoraproject.org/rpms/redhat-rpm-config/blob/master/f/redhat-rpm-config.spec#_118 22:27:28 YES!! ... thank you 22:27:29 pgreco has good memory 22:27:56 hehe 22:28:00 * King_InuYasha was one of those folks who pushed for those features in the first place upstream, so he remembers when they became a thing... 22:28:19 shame that we don't have %elif though 22:28:29 yeah, they are really useful 22:28:41 maybe for fedora 144 we'll have %elif 22:28:47 :) 22:28:47 nah, we have it in Fedora 33 now 22:28:57 it's a new feature in rpm 4.16 :) 22:29:40 nice, it should be backported to 4.14, so we can use it in epel 22:29:52 sorry rpm 4.15 22:30:00 so we've got it since Fedora 31 22:30:13 pgreco: sounds like a good contrib idea for centos stream 22:30:15 yeah, I want it backported to rpm 4.14 in el8 22:30:15 * carlwgeorge ducks 22:30:38 carlwgeorge: you're a big guy, you don't need to duck from me :D 22:30:51 although... this might run afoul of guidelines... 22:31:16 "All package dependencies (build-time or runtime, regular, weak or otherwise) MUST ALWAYS be satisfiable within the official Fedora repositories." 22:32:03 but this is clearly a good exception... so perhaps I should just not have brought it up. ;) 22:32:08 that's a rather annoying rule, actually 22:32:51 I mean as epel we can just add it to our exceptions to guidelines... we already have some 22:32:57 right 22:33:09 * King_InuYasha thinks that rule is a little too harsh even for Fedora 22:33:28 any example of where it doesn't work for Fedora? curious 22:33:37 I guess for weak dependencies that's a bit harsh 22:33:39 Correct, because there are plenty of weak dependencies that aren't in RHEL 22:34:35 I think it was designed to avoid fedora adding stuff depending on rpmfusion say... which kind of is an implicit endorsement of the name rpmfusion uses. 22:34:40 well, a weak dependency, that doesn't trigger an error if missing, it is satisfied (though I could argue it both ways) 22:35:50 weak dependencies can be done both ways though, so I guess rpmfusion can always add it on their side 22:35:54 anyhow, thats a sidetrack. I think this aproach is a good one. :) 22:35:57 "this package enhances Fedora package foo" 22:36:04 That's a good point, we could use recommends for the epel-next-release dependency 22:36:35 carlwgeorge: The recommends would need to be on centos-stream-release though 22:36:36 or require it if epel-release is installed? 22:37:00 if centos-stream-release, epel-release recommends epel-next-release 22:37:05 * cyberpear not a fan of bool deps 22:37:26 recommends is strong enough, yeah. at least for now. I wonder if anyone is working on the ability to not install it, like apt has 22:37:44 --setopt=install_weak_deps=False? 22:37:53 oh huh 22:38:02 if that's a thing our weak dependency guide is wrong 22:38:09 such an obtuse flag, we need a shorter version 22:38:14 King_InuYasha: that's normally part of my kickstarts :) 22:38:15 I've been begging for years 22:38:30 https://docs.fedoraproject.org/en-US/packaging-guidelines/WeakDependencies/ 22:38:39 King_InuYasha: want to document that workaround? 22:38:49 --install-weak-deps or --no-install-weak-deps would be better 22:38:49 this is plain wrong: "Future versions of dnf might also allow to switch weak deps off on the command line.)" 22:38:58 * King_InuYasha facepalms 22:39:06 agreed, but not being exposed as a separate option != doesn't exist 22:39:06 we've had the ability to turn it off since the beginning 22:39:41 the reason --excludeWeakDeps exists in kickstarts is because kkofler and I wrote that change to pykickstart :) 22:39:55 Anyway ... getting back from Fedora to EPEL ... carlwgeorge now that I have the right command, I'll get you the rpm right after the meeting. 22:40:18 King_InuYasha++ for kickstart work 22:40:53 carlwgeorge: Anything else before we move on? 22:40:53 if i understand the syntax right it should be `Recommends: (epel-next-release if centos-stream-release)`, added to epel-release 22:41:10 carlwgeorge: correct 22:41:13 * michel_slm will submit PR to update the Weak Dependencies doc 22:41:18 nope, that's it from me 22:41:26 thanks michel_slm 22:42:00 OK, up next is missing -devel packages 22:42:27 Actually, I think this one sorta fixed itself ... but I wanted to bring it up incase anyone had any comments. 22:43:06 * pgreco needs to read that thread again 22:43:07 I mean fixed itself, because the email thread sorta worked out, commands, spec file and all, how to do it. 22:43:51 needs a doc? 22:43:51 I was thinking of gathering that up and getting it into the EPEL wiki. 22:44:01 Ha ... ya beat me. :) 22:44:03 * nirik wasn't sure from the list thread what in the end worked or was made to work 22:44:57 Since I have three packages that never made it into CentOS Devel ... I can take one or two and work through the steps, making sure things work. 22:45:07 (btw, it's called centos-release-stream unless it's being renamed) 22:46:02 cyberpear: Thanks ... seems like an odd ordering of the name ... but ... if that's what it is. 22:46:48 there's lots of centos-release- packages, too 22:47:26 Hopefully I can get that done before the next meeting, and we can see if the -devel instructions are usable. 22:48:19 Anything else about missing -devel packages? 22:49:01 #topic EPEL-7 22:49:32 I don't have anything for EPEL-7 ... does anyone else? 22:49:38 nope 22:49:50 #topic EPEL-8 22:50:12 whoo boy 22:50:27 :) 22:50:46 I don't know what we're going to do about EPEL 8 22:51:02 what do we need to do? 22:51:27 cyberpear: it was renamed, but the old one is still what's in the linux repos for the conversion 22:51:31 well, at least by the end of next year, we need to transaction all mock configs off centos-8 22:52:02 yeah, but it's not clear to what... I think we will need to wait 6months or so and see what happens. 22:52:10 yeah 22:52:15 the options right now aren't great 22:52:19 True ... but I figure that gives us six months to look at what develops, and then 3 to 4 months to implement. 22:52:40 Ya'll type too fast. :) 22:54:17 besides that, I got nothing 22:54:40 Me either, or at least nothing we haven't already covered. 22:54:47 #topic General Issues / Open Floor 22:54:58 we'll just have epel-8-next / epel-next-8 and archive epel8 next year, right? 22:55:17 :) 22:55:26 ha. no. 22:55:49 I don't know about everyone else, but I'm not going to be around the next two fridays. 22:56:06 same 22:56:16 I'm OK to skip until January 22:56:17 I hadn't even considered that the meetings were happening 22:56:17 Is everyone ok with not having a meeting for two weeks, so the next will be Jan. 8, 2021 22:56:23 Jan 8? 22:56:24 yup 22:56:26 * nirik will be as much as he is today, which is to say... only if I happen to be at the keyboard. ;) 22:56:27 sounds good 22:56:34 yep 22:56:46 passed ... no meetings until Jan. 8. 22:56:50 omg, I can't believe I'm saying happy new year 22:57:11 can't be worse than 2020 22:57:14 :D 22:57:30 Happy New Year to everyone I don't see until then! 22:57:44 can we add a symlink for epel6 on archive? 22:58:04 I've always done yum install dl.fpo/pub/epel/epel-release-latest-6.noarch.rpm 22:58:49 I'd assume I could do the same w/ dl.fpo/pub/archive/epel, but it is not symlinked there... (I'll never remember that it's version 6 release 18 or whaever) 22:58:51 cyberpear: it's eol? 22:58:57 it is, but it's still useful 22:59:15 you'd be surprised how many were still using el5 this year 22:59:16 I was talking to someone yesterday and said something along the lines of "The year is finally over" ... but then we both quickly said "Nope ... 2020 isn't dead until it's dead. Something else can pop up in two weeks." 22:59:52 well, the way we are setting the link will mean that one has to be done totally seperately... 22:59:54 I mean, I was saying that last week 23:00:11 I guess we could, but seems a lot of work for something that is end of life and we don't want to advertize anymore 23:00:13 and then 2020 smacked me in the face with an awful reminder that, yes, 2020 is not over yet 23:00:50 cyberpear: if we did that, then people would keep using it 23:00:51 well, last week was an apt way to end the year 23:00:56 at least I asked... if it's too much effort, don't worry about it 23:00:57 cyberpear: can't you just hard code it? it's not ever going to change 23:01:11 pgreco: ugh 23:01:18 on CentOS 6, it's in extras... on RHEL 6, I always need the URL 23:01:57 https://dl.fedoraproject.org/pub/archive/epel/6/x86_64/Packages/e/epel-release-6-8.noarch.rpm 23:02:06 is there an EUS for rhel6? 23:02:12 yeah... thankfully it is linked 2 directories up 23:02:14 Thank you everyone for not only coming this week, but coming all the times pasts. It hasn't always been easy but we've kept EPEL stable, relavent and what people wanted. 23:02:16 ELS, for 3.5 more years 23:02:19 I think Evolution said it best, it's not truly stable till it's eol, no more of those pesky updates 23:02:29 :) 23:02:32 ha 23:02:43 hehehe, he's not wrong there 23:02:47 I'll talk to ya'll next year. 23:02:52 see ya! 23:02:55 #endmeeting