22:00:34 #startmeeting EPEL (2020-12-11) 22:00:34 Meeting started Fri Dec 11 22:00:34 2020 UTC. 22:00:34 This meeting is logged and archived in a public location. 22:00:34 The chair is tdawson. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:00:34 Useful Commands: #action #agreed #halp #info #idea #link #topic. 22:00:34 The meeting name has been set to 'epel_(2020-12-11)' 22:00:35 #meetingname epel 22:00:35 The meeting name has been set to 'epel' 22:00:36 #chair nirik tdawson bstinson pgreco carlwgeorge michel_slm 22:00:36 Current chairs: bstinson carlwgeorge michel_slm nirik pgreco tdawson 22:00:38 #topic aloha 22:01:07 .hello salimma 22:01:09 howdy 22:01:09 michel_slm: salimma 'Michel Alexandre Salim' 22:01:15 Hi michel_slm 22:01:18 Hi carlwgeorge 22:01:40 * michel_slm found a really cool site on Mastodon but forgot to re-toot it and now couldn't find the original toot to cite :p 22:01:42 hllo 22:01:49 search is hard 22:01:51 Hi smooge 22:02:00 first meeting in forever 22:02:31 smooge: Does this mean that grobisplitter is working again? 22:03:04 smooge: We definatly miss you and your wisdom of the ages 22:05:25 Small crowd today, but that's sorta expected this week. 22:05:36 #topic Old Business 22:05:43 EPEL-Packaging-SIG 22:05:49 oh why, is something big going on? >:) 22:06:14 wow, didn't expect this to get bumped over epel-next :) 22:06:31 Yep, apple is giving away free amazon kindles with every purchase of a google nest 22:06:33 tdawson, so I had to back grobisplitter back to the one we had previously 22:06:35 I think it's time to review the changes I proposed last week and in the mailing list a couple of days ago 22:07:05 (I also propose we consider the document live after today, and upcoming changes go into an EPEL ticket first) 22:07:42 I'm ok with that. I didn't see any negative things to the email. 22:07:45 tdawson: apple, amazon, google... get netflix and facebook on it and it's a perfect... er what's a Trifecta with 5 sides 22:08:07 I am I am ok 22:08:25 quintecta ? 22:08:54 pentagram 22:09:18 * michel_slm just realized this is logged and hoping HR didn't notice 22:09:26 :p 22:09:37 :) 22:09:58 Oh ... ya ... probrubly shouldn't have mentioned all those. 22:10:50 Anyway ... the changes you have to the SIG doc look ok to me. I know I'm ok with voting it in if others are. 22:11:04 I take it tdawson and smooge just voted yes. any others? 22:11:12 let me fetch the link to the email for completeness 22:11:55 I don't have a vote these days 22:12:31 https://lists.fedoraproject.org/archives/list/epel-devel@lists.fedoraproject.org/thread/X2K3IMNBYLVOHGW6XYECB2F3EX4C2SJS/ 22:12:48 i haven't managed to look over the whole doc, but what i've seen so far looks good. +1 22:12:52 this for the wiki changes - https://fedoraproject.org/w/index.php?title=EPEL%2FPackagers&type=revision&diff=596506&oldid=592132 22:13:40 smooge you are an emeritis member ... doesn't that mean you always get a vote? But ... either way, we'll put you down as a non-committe member voting positive 22:13:54 mark me down +1 22:15:12 #info EPEL Packagers SIG documentation voted on: Yes - 4 : No - 0 : Absent - 2 ... with 1 yes from a non-committee member. 22:15:45 yay! 22:15:57 :) Ya!! 22:16:14 I don't have anything else on this topic. was busy catching up on the CentOS news as well as... ah, even worse news for my employer :p 22:16:35 michel_slm: Sounds good. And thanks again for moving this forward 22:16:41 OK, next topic epel-next 22:17:29 carlwgeorge: How are things progressing? 22:18:36 smooge hooked me up with the grobisplit repos, i was able to implement the koji stuff in staging and did a successful build 22:18:58 next up is getting permissions to staging bodhi to figure out those steps https://pagure.io/releng/issue/9794#comment-705881 22:19:17 Cool ... it's really moving along. 22:19:29 if anyone is curious what that looked like https://koji.stg.fedoraproject.org/koji/buildinfo?buildID=1621291 22:21:01 * michel_slm would like to clarify that FB <3 Stream and our bad news is unrelated 22:21:37 Nice ... it looks like the repos are all setup and it's looking good. 22:22:03 after that it will be getting pungi and other thigns to produce and mirror manager and .. 22:23:13 All those ... little things? ... that make it usable. 22:24:04 Does it seem like it will at least be done in stg before the holidays? 22:24:53 I am not sure 22:25:03 releng next week will only be mohan 22:25:15 nirik is out until January 22:26:16 Well ... I guess if it's not all setup ... at least test packages can be built in stg. 22:26:46 carlwgeorge: Is there anything you need from us to help? perhaps do some test builds? 22:27:08 if you like, but i don't think it's necessary 22:27:37 so much of this is just reused and tweaked from epel8, so it's not uncharted territory 22:28:48 Sounds good then. anything else before we move on? 22:29:04 nope, go ahead 22:29:53 #topic EPEL-7 22:30:29 I don't have anything for EPEL7 ... does anyone else? 22:31:10 * tdawson takes a breath before moving to the next topic. 22:31:18 #topic EPEL-8 22:32:27 The biggest real discussion I've seen is what to do about missing -devel packages. 22:33:01 Most of the other discussions were misinformation, or emotions about the CentOS Linux news. 22:33:33 Although I did put out my proposal, I did have one question. 22:33:36 as someone bitten by this in the past... is there a procedure to follow that's not one-off fixes? I think one of the missing dependencies I have (might have been libuv-devel) ended up being shipped separately 22:34:16 (need to catch up on tdawson's proposal, apologies I missed it) 22:35:08 * nirik is here now... ca read back 22:35:16 That's sorta what my proposal is about. We just assume that Red Hat will not release those packages, and we figure out doing things compeltely within EPEL 22:35:59 bstinson: carlwgeorge: Do either of you know if Red Hat is going to allow CentOS Stream Devel to exist? Or at least be populated? 22:36:03 there is a procedure, but sadly it's package by package requests https://wiki.centos.org/FAQ/CentOS8/UnshippedPackages 22:36:09 I like Miro's proposal. if we can just take that unpublished -devel RPM and push it to epel instead of an official repo that would be sweet 22:36:11 the most correct way is to ask Red Hat for them to be shipped 22:36:35 * nirik hasn't read any of the proposals. were they on the list ? today? 22:36:39 no "building twice and hope it matches" and also no support burden for RH. though yes, having RH ship it officially is even better 22:36:40 tdawson: cs devel did exist, and we were required to empty it out 22:36:47 it's from 4 hours ago or so in epel-devel 22:37:01 carlwgeorge: sigh 22:37:09 ah, well, I am sure I will read it when I catch up on email. I will abstain on any voting on it until then tho. :) 22:37:22 i haven't read miro's proposal, but custom spec files to just ship a -devel package derived from a rhel spec file can be added to epel without review 22:37:31 i.e. there is a blanket review exception for them 22:37:36 nirik: Oh, it's definatly not in the "vote on it" stage. It's on the "is this possible" stage. 22:38:25 carlwgeorge: if that's the case I wonder how easy it is to automate - have a tool that takes a spec, rename it to -devel and strip all the other packages 22:38:43 michel_slm: carlwgeorge: Correct ... Miro thought there was something prohibiting them, but there isn't. I've got a half written reply ... but if others want to let him know he could do exactly what he proposed, they can beat me to it. 22:38:46 or... if Koji can be told to "use this spec, but filter out all but -devel" that would work too 22:40:33 The more I think of it ... writting scripts to do that to specs ... it actually sounds easier that fighting with grobisplitter and module specs. 22:41:22 it can't have the same name as the rhel package tho... 22:41:25 I like my second proposal better, but yeah scripting would mean not having to change Koji 22:41:39 nirik: yeah, I think the script will have to rename ${name} to ${name}-devel 22:42:20 yeah... as long as name doesn't have a -devel subpackage that would probibly work. 22:42:33 Ya .... I'd rather change grobisplitter than all the testing that would have to go into koji to do a change like that. 22:43:18 bullet 3 here https://docs.fedoraproject.org/en-US/packaging-guidelines/ReviewGuidelines/#_package_review_process 22:44:01 carlwgeorge: Ya!! 22:44:25 nice! 22:45:27 thinking out loud, the build steps can be kept, all the parts in %files for the other subpackages prepended with %exclude, the name and summary and BR/Req etc. from devel added to the top and hey presto, you have a derived -devel-only package 22:45:52 michel_slm: Yep ... my thoughts exactly. 22:45:54 so we can't just take th epackages out because they are in modules which have 'rules' which would not let you install them (aka filtered?) 22:46:04 sorry scroll stopped 22:47:57 smooge: I didn't quite follow what you meant. Were you saying that if the missing package was in a module, we wouldn't be able to do this? 22:48:01 we also have to be careful with looking at the existing modules to see if there are rules for basically conflicts with the devel modules. I ran into this a lot with the beta 22:48:25 with -devel rpm package (not devel modules) 22:49:07 basically modularity rules seem to be able to do 'you can't have this package installed on your system' and some of the -devel packages were in one or two of the modules long ago 22:49:23 Youch ... correct. 22:49:32 now grobisplitter removes that.. but joe-blow at home 22:49:46 I think libuv-devel might have been one of those packages. 22:49:52 sorry packager-blow at home trying to get that 22:50:15 working in his fedpkg local.. can't even if we have a module that says here it is 22:50:20 ah yeah, I remember libuv-devel took a long time to sort out 22:50:21 I thought that was moving into a rpm header... 22:50:33 they can with a mock config 22:50:39 shouldn't any modularity rule apply to modular RPMs only, and if it affects plain RPMs then it's a bug? 22:50:53 nope 22:50:55 I must sound really naive 22:51:09 because a module needs to be abel to replace a non-modular package 22:51:52 also because that metadata needs to persist in the repo (and RHEL keeps a ton of modular component history in its shipped repodata) 22:51:56 there are legitimate reasons for this 22:52:18 ah yes.. bstinson I had that problem with grobisplitter on centos-stream 22:52:44 patrick had put in a 'fix' to say never mind.. but it complained about missing metadata it couldn't check 22:53:15 or stephen g did .. not sure which 22:53:38 * bstinson looks at perl in a way as to suggest that it highlighted similar problems before :P 22:53:58 my understanding is that some of that metadata was going to move to being in the actual rpms, but I can't recall if thats happened yet or not 22:54:10 anyway.. if you make a module with these things.. you may also need to put every package that the replacing module has in it 22:54:11 definitely not yet in EL8 22:55:01 perhaps thats a 9 thing. 22:55:04 there is one metadata which has been put into the rpms but it is not related to this.. it is 'you can't install a modular rpm like it was a bare module.. rpm is to choke if you try and don't turn off a flag' 22:55:05 * nirik shrugs. 22:55:23 the hotpatch flag I think 22:55:48 the filter rules are done at a 'higher' level 22:56:17 Our time is running low, and we can continue to talk about this, but I just want to ask if there was anything else someone really wanted to bring up? 22:56:43 I think Neal is going to pop up soon and might want to say something 22:56:50 anyway.. from talking with the original modularity people at EPEL-8 standup for what could be done in EPEL-8 it was pretty much make a module with a different yaml than the Red Hat one which has all the items unfiltered 22:56:52 * nirik had a short item for open floor, whenever that is 22:56:54 * King_InuYasha waves 22:56:58 .hello ngompa 22:56:59 King_InuYasha: ngompa 'Neal Gompa' 22:57:08 it's been a rough week, sorry about being late 22:57:09 #topic General Issues / Open Floor 22:57:18 Hi King_InuYasha 22:57:25 ok nirik has first dibs on open floor? 22:57:30 yep 22:57:49 #info user managed side-tags are now available in epel7/8 and bootstrap mode is enabled (announcement to come later today with details about that) 22:58:04 Ya!! 22:58:05 nirik++ 22:58:05 michel_slm: Karma for kevin changed to 16 (for the current release cycle): https://badges.fedoraproject.org/tags/cookie/any 22:58:24 we need to implement karma for EL release cycles :p 22:58:28 woot! 22:58:34 nirik++ 22:58:51 nifty. nirik++ 22:58:51 orionp: Karma for kevin changed to 17 (for the current release cycle): https://badges.fedoraproject.org/tags/cookie/any 22:59:34 King_InuYasha: Did you have something too? 22:59:52 not really, I just freaked when I realized that I missed everything 22:59:57 :) 23:00:21 my day started at 4am EST 23:00:34 the tail end of a hackathon 23:00:52 Well, looks like our time is up. Thank you everyone for keeping this civil and on topic. I know that was harder this week and I appreciate it. 23:00:54 same here... koji outage pages woke me. ;( 23:01:21 man, I feel for you nirik 23:01:32 thanks tdawson 23:01:33 Talk to you all next week. 23:01:34 last week, I was on-call at work 23:01:40 * smooge can't sleep the clowns keep eating me 23:01:41 #endmeeting