20:00:46 #startmeeting EPEL (2021-06-16) 20:00:46 Meeting started Wed Jun 16 20:00:46 2021 UTC. 20:00:46 This meeting is logged and archived in a public location. 20:00:46 The chair is tdawson. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:46 Useful Commands: #action #agreed #halp #info #idea #link #topic. 20:00:46 The meeting name has been set to 'epel_(2021-06-16)' 20:00:50 #meetingname epel 20:00:50 The meeting name has been set to 'epel' 20:00:54 #chair nirik tdawson bstinson pgreco carlwgeorge michel_slm 20:00:54 Current chairs: bstinson carlwgeorge michel_slm nirik pgreco tdawson 20:00:59 #topic aloha 20:01:21 .hi 20:01:23 dcavalca: dcavalca 'Davide Cavalca' 20:01:28 Hi dcavalca 20:01:31 howdy 20:01:34 morning 20:01:48 Hi carlwgeorge 20:01:55 Hi nirik 20:02:09 hey everybody 20:02:21 dcavalca: hello mr board member!!! 20:02:53 Hi pgreco 20:03:14 dcavalca: Congratulations 20:03:49 lol thanks 20:04:14 Radek: packages are not branched by default for EPEL, if something is missing please file a bug requesting it 20:04:15 .hello salimma 20:04:15 michel: salimma 'Michel Alexandre Salim' 20:04:19 * michel notes that Davide does not like being called Il Direttore 20:04:20 hi all! 20:04:54 ha, having a username that is just my first name really causes confusion with the IRC->Matrix bridge 20:05:03 Hi michel 20:06:01 #topic Old Business 20:06:14 Let's start with the EPEL packaging SIG 20:06:16 * nirik always thinks tdawson is making fun of him with that. 20:06:33 nirik: with what? 20:06:38 Any new business? 20:06:42 ok, so I looked into how to create a template 20:06:43 "old" business. :) 20:06:51 * nirik is old. 20:06:55 it looks... a bit involved. anyone knows which repo the template is in? 20:07:18 nirik Are you older than me? or smooge? 20:07:25 but anyway, I'm thinking the use case we need to optimize for is if you want package A in EPEL, and it has 5 other dependencies... and you want to streamline that process 20:07:30 * michel will turn 42 later this year 20:07:31 he is older than me 20:07:50 nirik I don't think you're even older than me 20:07:53 so I guess the most meaningful question in the universe is "how old is Michel" 20:08:11 michel: Ya, that's always a pain when you need to do alot of dependencies. 20:08:17 #info Bugzilla template instructions: https://bugzilla.readthedocs.io/en/latest/integrating/templates.html 20:08:29 but barely 20:08:34 I'm so old I forget how old I am. ;) 20:08:42 tdawson: so... I just started using Mutt with anger on my CentOS box, and have what I think is a perfect use case for this tool I want to write 20:08:47 your old enough to enjoy your applesauce 20:08:57 goal: feed it a yaml (or another markup) describing the dependency tree 20:09:31 it will then autofile the bugs and link them correctly for you. so the template can just be maintained in the repo for this tool, and you don't need to even customize it, it will be prefilled 20:09:48 That will be cool 20:10:15 (mutt is in EPEL, but to auto-complete contacts it needs two other tools, khard and vdirsyncer, and they have about 7 Python deps between them) 20:10:58 I can probably add a way to check on stale requests there too and automate the filing of infra requests next. but yeah, going to dogfood it for this set of packages 20:11:14 * nirik has to go move water, back in a min. 20:11:16 that's it from me - just finished my second-to-last oncall before i can focus on OSS :) 20:11:45 oh, once I have something working, any preference for where the code should live? do we have a group on Pagure? 20:11:49 That will be nice. I hope the dogfood tastes good ... or works, whatever the proper phrase is 20:12:05 * michel tried cat food once. it's not terible 20:12:09 * michel * tried cat food once. it's not terrible 20:13:01 I don't know if epel is a group - https://pagure.io/epel 20:13:22 But we do have a plac eto put files 20:13:32 Huh ... spelled right, spaced wrong 20:13:58 place to put files 20:14:21 that seems a repo 20:14:30 I can put it on my pagure / github for now 20:14:42 oh fun it has the old board members listed still :D 20:15:49 Yep ... need to be older than 50 to be on that list 20:15:58 Oh ... you meant something else. :) 20:16:04 * michel laughs 20:16:28 I'm charting my age by now many "Over X support groups" I can join at work 20:16:29 you could make an epel group 20:16:43 and then start setting things up that way 20:16:43 we grouch about how the young 'uns don't get Matrix references anymore 20:16:44 and Hackers 20:16:56 that's what we did in fedora-workstation 20:17:00 won't the epel group clash with the epel repo? 20:17:05 nope 20:17:19 see https://pagure.io/group/fedora-workstation 20:17:25 vs https://pagure.io/fedora-workstation 20:17:25 oh 20:17:31 Nice 20:17:38 ok, should I do that? 20:17:47 yes 20:17:58 we can vote on it if people want but I forgot how to propose :p 20:18:03 once you have an epel group, you can make projects in the namespace 20:18:12 and you can make the epel repo as part of that group too 20:18:25 can we rehome it, or do we need to create a new one? 20:18:27 (even though it's not in the namespace, because pagure considers namespace and groups separate) 20:18:32 ah ok 20:18:46 it will be in the group but still at the old URL 20:18:46 yep 20:18:55 namespaces are created with groups, but projects don't have to be in the namespace 20:18:57 kind of like... Matrix Spaces /ducks 20:19:10 it's weird, but that quirk exists because pagure was designed with dist-git use-cases in mind 20:19:20 *nods 20:19:22 * nirik is fine either way. For something like this it seems like it would be fine to just put it in epel project, but whatever. 20:19:23 as an admin, you can create namespaces with no groups 20:19:27 so all projects have to be accessible at /rpms/pkgname 20:19:30 yes 20:19:55 it's useful because namespaces don't wind up being tied to group permissions either 20:20:03 which as far as I know, nothing else does 20:20:08 * nirik is now getting confused what you are talking about 20:20:08 nirik: it might be flexible enough to cover non-epel use cases, and... we might want a separate docs repo at some point. so, yeah, why not 20:20:32 nirik: I'm just explaining how groups and namespaces work in pagure to Michel Alexandre Salim 20:20:42 I think we're ratholing a bit about how namespaces work, we can move on 20:20:51 ok, but you are talking about pagure.io and a epel namespace, not src.fedoraproject.org right? 20:20:59 OK, anything else for the SIG? 20:21:35 Let's move on to epel-next 20:21:37 nirik: yes 20:21:40 nirik: yes, but I guess they're both Pagure underneath so the same capabilities exist 20:21:44 carlwgeorge: Any new news? 20:22:03 well, yeah, but src is different for groups... but anyhow 20:22:07 carlwgeorge: I forgot to test the new repo but I can do that now if needed 20:22:45 #info epel next open for testing https://lists.fedoraproject.org/archives/list/epel-devel@lists.fedoraproject.org/message/6VAHYUU3Z7DYG6W6UBYRN3QSCDOG73TM/ 20:22:58 Ya!! 20:23:08 cool 20:23:17 How long do we plan on testing it? 20:23:20 * michel installing now 20:23:24 i figured a few weeks 20:23:26 if I get disconnected blame carl 20:23:28 * dcavalca already added it to the internal repo slate 20:23:59 i'd love help from people in the kde sig to request epel8-next branches 20:24:07 I just got my last qt5 branches, so that will give me time to get at least the qt5 updates in there. 20:24:37 i've been doing a few scratch builds, and cherry-picking cmake fixes from rawhide to epel8 branches so they're ready (would be needed anyways since 8.4 for regular epel8 builds) 20:25:29 in related news, the fedpkg pr was finally merged upstream https://pagure.io/fedpkg/pull-request/429 (we weren't waiting for it, but good to see) 20:25:30 carlwgeorge: I've got a little script for the kde branches, now that I have all my qt5 branches, I'm going to do the other 350 20:25:41 Ya!! 20:26:37 doesn't break anything but some Qt packages are still missing right now 20:27:17 tdawson: woot! 20:27:35 ps my use case is I have the nextcloud client installed 20:27:53 * pgreco updates epel-release and sees epel-next-release being pulled :D 20:27:56 if this is ready for others to contribute to, I can see about branching the right packages 20:28:11 nextcloud uses qt5 ? 20:28:19 tdawson: yup 20:28:20 yes please, request branches 20:28:37 the desktop client I assume? 20:28:50 the desktop client 20:28:52 also please let me know if https://fedoraproject.org/wiki/EPEL_Next#Example_Workflow is missing any steps 20:28:56 Huh ... well, I got the branches litterally minutes ago, and I plan on doing the builds this evening ... hopefully by tomorrow we'll have it in -testing 20:29:44 tdawson: is epel-packagers-sig in the ACL or just you? but I can wait until tomorrow, let's not step on each other's toes 20:30:05 michel: It's the kde-sig that's in the ACL, of which I'm a part of. 20:31:02 ah. I might want to join that. hm, would nextcloud-client be in scope as it's a Qt,not KDE app? 20:31:12 carlwgeorge: I use fedpkg instead of git (fedpkg switch-branch epel8-next, fedpkg merge, fedpkg push) 20:31:37 .whoowns qt5-qtenginio 20:31:37 I can ask nextcloud-client to be branched, since it will need to be rebuilt 20:31:38 pgreco: owner: rdieter; admin: lupinix 20:32:02 carlwgeorge: that's another one that needs to be rebuilt for epel-next 20:32:13 michel: if rex is the owner, I'd say it's a kde thing. 20:32:18 yeah i've got a big list from repoclosure 20:32:28 great 20:32:36 .whoowns nextcloud-client 20:32:38 michel: owner: germano; admin: nonamedotc; commit: taaem 20:32:42 that's the only one that was failing in my wife's laptop 20:32:47 it's germano, nonamedotc and taaem 20:32:56 i blame Eighth_Doctor, he's the one that asked for qt to be rebased 20:33:19 blaiming Eighth_Doctor is always a good way to go 20:33:46 Conan Kudo: move fast and break things :p 20:33:53 .... 20:33:57 I hate you so much now 😛 20:34:24 🕺 20:34:57 Oh, I've already done qt5-qtenginio ... but I haven't done the update yet, I was going to do all the qt5 updates on one update... which, maybe wasn't the best idea, but saves me from having hundreds of updates at once. 20:35:08 there are probably people who do really think I think like that too... 20:35:15 Anyway, qt5 packages should be in -testing by tomorrow, hopefully. 20:35:15 between qt in 8.5 and cmake in 8.4, nothing but love for Eighth_Doctor's work 20:35:23 *laughs* 20:36:09 * Eighth_Doctor wonders when he started to care about things in RHEL so much... 20:36:10 Anyway, anything else for -next ? 20:36:28 nope, we can move on 20:36:37 OK, moving on. 20:36:46 Badges and Logos 20:36:47 any idea when we'll have it for cs9? 20:37:07 Issues opened for both 20:37:09 i plan to start epel9-next as soon as we're "done" with epel8-next 20:37:17 epel commitee badge - https://pagure.io/fedora-badges/issue/829 20:37:20 i.e. pushed to stable and announced 20:37:30 epel logo - https://pagure.io/design/issue/770 20:37:39 👍️ 20:38:05 ooo badge! 20:38:06 https://pagure.io/design/issue/770#comment-737519 is the one you want :D 20:38:08 The centaur might get pushed to be a mascot instead of a logo 20:38:37 I filed https://bugzilla.redhat.com/show_bug.cgi?id=1972910 for nextcloud-client, please correct me if I wrote anything wrong in the description 20:39:23 I like the ants. Look like... Space Ants with MBAs 20:39:30 nextcloud-client is blocked by qt5-qtwebengine and kf5-kxmlgui, according to my scratch builds 20:40:31 michel: Just curious why you didn't just ask the maintainer to branch and build it on epel8-next, looks like they are quite responsive. 20:40:51 They've updated the package twice this year in epel8. 20:40:57 I can do that 20:41:03 are the dependencies ready? 20:41:24 nope 20:41:56 Actually, yes they are, I have all the qt5 packages I've built in override 20:42:02 neat 20:42:05 aha 20:42:16 And looking at it's dependencies, it doesn't need any of the missing ones. 20:42:58 Anyway, back to badges and logo's. If anyone has any ideas or comments, feel free to put them on the issues. 20:43:39 We're getting low on time, I'm going to run through the usual topics. 20:43:45 #topic EPEL-7 20:43:53 Anything for epel7? 20:44:03 I think carlwgeorge solved the only thing I had 20:44:15 yeah i retired pidgin-epel 20:44:41 #info retired pidgin-epel from epel7 https://bugzilla.redhat.com/show_bug.cgi?id=1972323 20:44:52 glad we don't need it anymore 20:45:02 haven't since 7.3 lol 20:45:37 Yep, just that nobody noticed. 20:45:57 #topic EPEL-8 20:46:05 Anything for epel 8? 20:46:33 i'm sure there a bunch of those, where the rhel maintainer set the version-release higher, but there is potential for issues if the epel package is upgraded 20:47:11 wait... what's the context? 20:47:22 once a package is in RHEL we're supposed to retire it from EPEL, right? 20:47:35 correct, by policy 20:47:47 And technically, pidgin was retired 20:47:58 but sometimes the retire package has higher ENVR than the one included in RHEL 20:48:14 and that creates "issues" 20:48:24 well, that was a different case... that was pidgeon-epel 20:48:25 But we created a pidgin-epel, because the pidgin in RHEL didn't really have "pidgin" the binary. 20:48:33 it was missing some libs 20:48:38 (I thought) 20:48:48 the rhel package was just libpurple 20:48:49 pgreco: ah 20:49:04 Stream will hopefully make that easier, less opacity 20:49:05 until 7.3 when the rest of the subpackages were added 20:49:20 hm, though we still won't know which get pulled into the stable releases 20:49:56 it amazes me how schizoid RHEL is at times 20:50:43 Ya ... I'm not even going to comment on that cuz it gives me a headache. 20:51:00 Anyway, sounds like there aren't any epel8 stuff, so moving on to ... 20:51:06 #topic General Issues / Open Floor 20:51:34 I wanted to bring up a epel-release bug and ask folks to add comments to it if they care to: 20:51:50 https://bugzilla.redhat.com/show_bug.cgi?id=1969500 20:52:13 I looked at that. I'm for changing it to just 8 20:52:16 basically people setting releasever to 8.4 or whatever breaks epel 20:52:38 we could do that, or we could make the minor numbers go to the archive for that release... 20:52:39 i'm against changing it to just 8 20:52:41 I think the archived 8.1, 8.2, etc ... should be a seperate issue to discuss and figure out 20:53:01 when someone locks to an old minor release, not every epel8 package will install 20:53:11 this is a bad idea 20:53:20 right, and no updates security or otherwise 20:53:38 i'm in favor of nirik's suggestion of adding the snapshotted epel8 archives to mirrormanager, to respond with compatible packages 20:53:41 we deliberately switched to $releasever because of the minor version based snapshots 20:53:49 old minor releases get security updates for RHEL only, right? 20:53:55 yeah, EPEL is frozen 20:54:14 michel: if and only if you're using eus proper, not a local frozen snapshot 20:54:14 ah. EPEL also got snapshotted? 20:54:33 indeed https://dl.fedoraproject.org/pub/archive/epel/ 20:54:50 yeah, it's a tradeoff for someone who sets it to say 8.2... they would get packages that would work with 8.2, but not the latest. 20:55:02 just today i noticed some directory snafu with the 8.3 archive, and mboddu got it fixed quickly https://pagure.io/fedora-infrastructure/issue/10041 20:55:12 and also keeping the $releasever may be useful to "downgrade from snapshot" testing 20:55:42 I worry that they would expect epel to behave just like rhel there tho... keep updating things in the y stream 20:56:07 that seems like a lesser problem than "this won't install" 20:56:15 it can't be affecting too many people... this is the first bug I have ever seen on it. ;) 20:56:43 i checked with my old work, where i deployed rhui right before i left. they have seen it too, just never complained. 20:56:48 people being on old point releases basically should suffer for their sins 20:56:49 That's what I worry about to. People start thinking that we build things for 8.2, 8.3 and so forth, when we don't. 20:57:00 thankfully they finally stopped defaulting customers to eus and leaving them there until they complained 20:57:42 Eighth_Doctor++ :) 20:57:47 i prioritize uninstallable packages over "why isn't this getting updates anymore", but that's just my opinion 20:57:52 symlink to 8 is not an option? 20:58:05 if someone wants the snapshot, they should manually select it 20:58:22 maybe it won't install, but it won't be outdated by definition 20:58:51 that is the question at hand, if they override their releasever to an old minor instead of just 8, do they get current epel (some uninstallable packages) or snapshoted epel at the minor they asked for 20:59:33 I'd go for current (and maybe failing) packages 20:59:37 pgreco: you mean alias 8 and 8.4 to current... and snapshots to older? 21:00:03 I'd symlink anything to current 21:00:04 right, 8.4 has no snapshot, so that should be a symlink to 8 21:00:12 if you want the snapshot, get it manually 21:00:35 but for older releases I'm torn 21:00:37 but then you need to edit repo files and hardcode the snapshot ? 21:00:44 yes 21:01:01 thats kind of a pain if you just want one or a few things from it. 21:01:41 I'm not sure what's best, tbf, but I don't like the idea of transparently offering something outdated 21:01:48 i say we think about it for a week, and vote then 21:01:55 on the Fedora side.. we make people who insist on old releases jump through hoops, right? 21:01:59 to switch to the archive repo 21:01:59 folks can add their thoughts to the bug also 21:02:06 michel: nope 21:02:23 michel: No, if you stay on fedora 15, you still have access to all of Fedora 15, no change necessary. 21:02:28 mirrormanager returns the thing you ask for, even if it's old 21:02:29 I think CentOS does, but fedora just has mirrormanager send them to archive 21:02:37 Don't know why I picked 15 ... it's 14 that I have a machine on. 21:02:46 MirrorManager is awesome like that 21:02:47 oh huh, didn't realize mirrormanager does that 21:02:51 I wish CentOS used it 21:02:56 * michel built some archive-handling code for nothing 21:02:56 it would make my life so much easier 21:02:58 trying for 9 21:03:01 Oh ... just looked at the time 21:03:17 Yep, let's think/talk about this during the week, and get back for a vote next week. 21:03:28 +1 21:03:36 yeah 21:03:43 see you next week, thanks tdawson 21:03:46 Thanks to everyone for coming, and for the good discussions. 21:03:54 #endmeeting