18:02:09 <smooge> #startmeeting EPEL (2019-10-16)
18:02:09 <zodbot> Meeting started Wed Oct 23 18:02:09 2019 UTC.
18:02:09 <zodbot> This meeting is logged and archived in a public location.
18:02:09 <zodbot> The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:02:09 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:02:09 <zodbot> The meeting name has been set to 'epel_(2019-10-16)'
18:02:09 <smooge> #meetingname epel
18:02:09 <zodbot> The meeting name has been set to 'epel'
18:02:09 <smooge> #topic aloha
18:02:09 <smooge> #chair nirik smooge tdawson bstinson Evolution pgreco sgallagh
18:02:09 <zodbot> Current chairs: Evolution bstinson nirik pgreco sgallagh smooge tdawson
18:02:09 <smooge> #topic Intros
18:02:27 <bstinson> yay another wednesday!
18:02:34 <pgreco> hello
18:02:41 <smooge> monday #3
18:02:56 <smooge> nirik is out. Evolution is in meetings
18:03:37 <smooge> #info Meeting is run from https://board.net/p/epel
18:03:37 <smooge> #topic Old Business
18:04:02 <smooge> ok I put out the email listing the 138 tickets open which seem to be epel-8 related
18:04:11 <sgallagh> .hello2
18:04:12 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
18:04:28 <smooge> hey sgallagh I wasn't sure yuou were around
18:04:49 <sgallagh> I was running a bit late. Here now.
18:04:58 <smooge> np
18:05:28 <smooge> #info Need to look out why some packages can't be branched (rsh)
18:05:49 <smooge> I think that is in pdc
18:06:05 <smooge> but I am not sure and I am not sure if it is possible to over-ride
18:07:19 <smooge> any other old business?
18:08:43 <smooge> #info Getting EPEL docs into docs.fedoraproject.org did not go smoothly
18:09:21 <smooge> I followed what pbokoc had said.. but it doesn't work well for a repository which is also getting lots of tickets as every person in fedora-docs got cc'd on tickets etc
18:09:39 <smooge> so I made pbokoc an admin for the time being until we work out what is needed
18:09:42 <smooge> #topic EPEL-6
18:09:43 <smooge> #info EPEL-6 is End of Life in 2020-11. It will be moved to archives in 2020-12
18:10:21 <smooge> #topic EPEL-7
18:10:21 <smooge> #info no news is good news
18:10:21 <smooge> #topic EPEL-8
18:10:21 <smooge> #info sgallagh says...
18:10:57 <smooge> how goes stuff sgallagh ?
18:11:20 <sgallagh> We're a little behind schedule, but moving forwards
18:12:01 <sgallagh> I created the module defaults branches for EPEL 8 and playground today
18:12:10 <sgallagh> Merlin is going to take over landing the remaining bits to start doing builds.
18:12:30 <sgallagh> My goal for this upcoming week is to do at least one successful module build against the epel8 platform
18:12:59 <sgallagh> My goal for the week to follow will be to try to generate an Uber-module that contains all the stuff that was filtered out of RHEL 8
18:13:21 <sgallagh> Though I expect that may extend to two weeks
18:13:33 <smooge> cool
18:14:03 <smooge> we will need to figure out how to get 'branches' from git.centos.org for that software
18:15:07 <sgallagh> Yeah, that's going to be fun
18:17:00 <smooge> bstinson, do you have suggestions?
18:17:25 <tdawson> Do you need any testers for modules?
18:17:41 <bstinson> we still need to discuss that a little bit on the CentOS side
18:17:45 <sgallagh> tdawson: I will definitely ask for them as soon as I have a successful build
18:17:55 <tdawson> ok
18:19:00 <sgallagh> smooge: In the short term, we *can* direct MBS to point at a non-default git repository for the components
18:19:13 <sgallagh> So I will perhaps do that until the syncing/merging is worked out
18:19:30 <sgallagh> Sorry, I should probably clarify that
18:19:34 <smooge> sounds good. I didn't know that
18:20:14 <bstinson> is epel wanting to consume that module directly from a repo that CentOS provides?
18:20:14 <sgallagh> The components section in the modulemd metadata has an optional `repository` field that can point to another repo: https://github.com/fedora-modularity/libmodulemd/blob/master/spec.v2.yaml#L263
18:22:00 <smooge> bstinson, sgallagh that might be one way to do it. if centos produced such a module in a repo we could sync and add to koji?
18:22:46 <sgallagh> smooge: Sorry, I meant that we could create a module in Fedora dist-git (modules/ prefix) whose components might be coming from branches in centos dist-git
18:22:52 <sgallagh> Such as for the Uber-module
18:23:11 <sgallagh> (Aside: any good ideas for what to call it?)
18:23:29 <tdawson> I like Uber-module ... that sounds good.
18:23:55 <tdawson> But ... it also doesn't really say what it has in it ... it just sounds cool
18:24:19 <sgallagh> Right, its purpose is to carry any of the subpackages that RHEL 8 didn't deliver.
18:24:27 <sgallagh> Like the libuv-devel package (for example)
18:24:43 <bstinson> if we can accelerate the necessary conversations in CentOS, i'd rather curate that module over on that side
18:24:58 <smooge> could someone take over the meeting. I need to do another meeting all of a sudden
18:24:59 <bstinson> and do it once rather than try to keep 2 of them in sync (we're considering shipping something similar)
18:26:08 <sgallagh> OK, I just spoke to an MBS developer and Fedora's MBS in prod currently disallows overriding the repository (for reproducibility requirements)
18:26:26 <sgallagh> I could ask them to add a whitelist, but if we could get the merged git sooner, that would be much nicer.
18:27:03 <tdawson> How about codeready-missing-module
18:27:29 <smooge> code-not-ready-module
18:28:11 <sgallagh> smooge: Too snarky, even for me :)
18:28:25 <smooge> sgallagh, yeah sorry
18:28:28 <sgallagh> tdawson: That could work
18:28:50 <bstinson> we should steer away from the codeready branding
18:29:00 <smooge> actually we should .. what bstinson said
18:29:17 <sgallagh> good point
18:29:21 <smooge> additional-powertools-module
18:29:35 <sgallagh> actually, maybe just `epel-platform`
18:29:36 <tdawson> We could shorten it to   missing-module
18:29:51 <sgallagh> With streams for each minor release
18:29:57 <smooge> not-the-module-you-are-looking-for
18:30:03 <tdawson> *laughs*
18:30:05 <sgallagh> Move along
18:30:14 <smooge> move-along-module
18:30:29 <bstinson> is there a reason that centos couldn't ship a repo with this module in place?
18:30:38 <bstinson> i *really* *really* don't want to do this twice
18:30:50 <smooge> agreed. because you will have to update it a lot
18:32:27 <sgallagh> bstinson: Actually, that might indeed be preferable
18:32:55 <sgallagh> It would be best if the CentOS folks could take ownership of automating it so that it's updated when RHEL releases errata
18:33:03 <sgallagh> (If they get out of sync, bad things(tm) might happen)
18:33:16 <pgreco> well, I was going to ask about that
18:33:19 <smooge> bad things being a broken build systme
18:33:27 <sgallagh> right
18:33:31 <smooge> oooh speaking of broken build systems
18:33:36 <sgallagh> ...
18:33:41 <pgreco> is the idea to use a split repo (rhel + missing parts coming from centos)
18:33:56 <pgreco> or is it easier just to switch it all to build from centos?
18:34:14 <smooge> #info EPEL8 was broken earlier this week because epel-rpm-macros got pushed with a change which should not have caused problems but did. We reverted until we figure it out
18:34:28 <sgallagh> Theory, meet practice...
18:35:44 <smooge> pgreco, so in theory.. if all the items are in a module, then it doesn't matter if we mix the streams as long as we have said module with the right metadata and Ursa prime does its stuff.
18:36:16 <smooge> if they aren't in a module then it gets hairy like when we tried to use the centos aarch64 7.7
18:36:23 <smooge> however...
18:36:54 <smooge> practice will show where our spherical cows grew udders and lost their aerodynamics
18:37:08 <sgallagh> bstinson: Does CentOS do the same filtering of subpackages, or does it actually carry them?
18:37:28 <tdawson> It filters them the same ... so it's missing the same packages.
18:37:30 <bstinson> we do the exact same filtering
18:37:30 <sgallagh> Because if they're available from CentOS, I think just using that as the buildroot makes more sense than RHEL
18:37:32 <sgallagh> damn
18:37:35 <smooge> sgallagh, currently centos filters because the modules defs are defined in the git
18:38:49 <smooge> sgallagh, but could something like how virt : virt-devel are kept in sync work?
18:39:42 <sgallagh> smooge: In theory, but someone actually has to do that work
18:39:55 <smooge> well you volunteered yourself earlier
18:40:06 <sgallagh> I volunteered to get us off the ground.
18:40:07 <smooge> so I check marked that off already
18:40:19 <sgallagh> I am *not* (and cannot) be a long-term maintainer
18:40:26 <smooge> understood :)
18:40:53 <bstinson> let's start with a discussion on centos-devel to see what that community wants out of missing -devel packages
18:40:58 * bstinson takes that action
18:41:49 <sgallagh> Thanks
18:43:05 <smooge> sgallagh, thinking about it the 'uber-module' would have had to build and include both sets of packages like libblueray and libblueray-devel
18:44:13 <sgallagh> It would have to build them
18:44:14 <smooge> I am trying to figure out if it would make sense to rebuild them or have a fake module like the way RHEL-8 does perl
18:44:16 <sgallagh> Not necessarily ship them
18:44:35 <sgallagh> smooge: Do we have access to them somewhere?
18:44:41 <smooge> not me
18:44:45 <sgallagh> The binary packages need to exist.
18:44:52 <pgreco> sgallagh, how is this affected by the module name difference between rhel and centos?
18:45:15 <smooge> sgallagh, the reason I was wondering was it would need to ship them since libblueray is not in a module
18:45:19 <sgallagh> pgreco: What module name difference?
18:45:39 <pgreco> el8_module+++++
18:45:49 <sgallagh> oh, hmm
18:46:13 <sgallagh> We might have some issues with %{release} in dependency fields :-/
18:46:21 <smooge> so if libblueray-devel was in a module.. would we end up with one of those 'oh your module has this thing we have to replace your non-module with it'
18:48:08 <smooge> anyway..
18:48:30 <smooge> bstinson thanks for opening an email thread. we will look for it on centos-devel and work from there?
18:49:12 <bstinson> yeah, let's start there
18:49:15 <smooge> #info bstinson will open a thread on centos-devel on how to deal with shipping non-shipped packages
18:49:35 <smooge> #info devconf.cz CfP ends in 7 days.
18:49:39 <sgallagh> smooge: I think I need to ponder that more deeply
18:49:46 <smooge> sgallagh, understood
18:50:33 <smooge> #topic open floor
18:50:59 * sgallagh falls through
18:51:51 <smooge> it is a deep pit filled with alligators for a reason
18:54:49 <smooge> ok nothing for open floor? The piranha are starting to get hungry so I am going to close this out.
18:55:57 <smooge> #endmeeting