17:00:25 <decathorpe> #startmeeting FESCo (2023-06-13)
17:00:25 <zodbot> Meeting started Tue Jun 13 17:00:25 2023 UTC.
17:00:25 <zodbot> This meeting is logged and archived in a public location.
17:00:25 <zodbot> The chair is decathorpe. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
17:00:25 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:00:25 <zodbot> The meeting name has been set to 'fesco_(2023-06-13)'
17:00:33 <decathorpe> #meetingname fesco
17:00:33 <zodbot> The meeting name has been set to 'fesco'
17:00:46 <mhroncok> .hello churchyard
17:00:47 <zodbot> mhroncok: churchyard 'Miro Hrončok' <mhroncok@redhat.com>
17:00:49 <decathorpe> #chair nirik, decathorpe, zbyszek, sgallagh, mhroncok, dcantrell, music, mhayden, Conan_Kudo, Pharaoh_Atem, Son_Goku, King_InuYasha, Sir_Gallantmon, Eighth_Doctor, tstellar
17:00:49 <zodbot> Current chairs: Conan_Kudo Eighth_Doctor King_InuYasha Pharaoh_Atem Sir_Gallantmon Son_Goku dcantrell decathorpe mhayden mhroncok music nirik sgallagh tstellar zbyszek
17:01:05 <decathorpe> #topic Init Process
17:01:13 <nirik> morning everyone.
17:01:25 <decathorpe> 'evening everyone  o/
17:01:58 <jvanek> hi all!
17:02:23 <decathorpe> hi! thanks for attending.
17:02:47 <tstellar> .hello tstellar
17:02:48 <zodbot> tstellar: tstellar 'Tom Stellard' <tstellar@redhat.com>
17:02:58 <zbyszek> .hello2
17:02:59 <zodbot> zbyszek: zbyszek 'Zbigniew Jędrzejewski-Szmek' <zbyszek@in.waw.pl>
17:04:06 <mhroncok> tstellar: congrats ;)
17:04:18 <decathorpe> welcome (back?) to FESCo :)
17:04:48 <zbyszek[m]> Hi Tom!
17:05:14 <decathorpe> ok, we are 5 (with sgallagh being in another meeting, so 5.5), so we have quorum
17:05:21 <tstellar> mhroncok: Thanks.
17:05:45 <sgallagh> Looks like my other meeting is not happening after all.
17:05:47 <sgallagh> .hi
17:05:50 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
17:05:55 <decathorpe> oh, even better
17:06:03 <decathorpe> #link https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/2Q7NDMZEQS5L4UZNTLL2HOB3ZE5P75RH/ Schedule
17:06:05 <sgallagh> (The other participant didn't turn up)
17:06:59 <decathorpe> #topic #3008 Change: Build JDKs once, repack everywhere
17:07:07 <decathorpe> .fesco 3008
17:07:08 <zodbot> decathorpe: Issue #3008: Change: Build JDKs once, repack everywhere - fesco - Pagure.io - https://pagure.io/fesco/issue/3008
17:08:07 <sgallagh> I know I'm going to get flack for asking this, but how badly do we care about the "Java" certification for Fedora (RHEL is a different story)?
17:08:22 <decathorpe> Ok, I'm not really sure where we stand here - there have been only two (partially) negative votes so far.
17:08:38 <decathorpe> Stephen Gallagher: that is a good question IMO
17:08:54 <jvanek> answered few times
17:09:05 <sgallagh> I mean, a package that walks, talks, quacks, flies and swims like a duck, but called a mallard, is still the same animal
17:09:23 <jvanek> unless oracle lawyer shot it
17:09:41 <sgallagh> That metaphor kind of got away from me, but I think you got what I mean
17:09:56 <decathorpe> even worse question: do we actually need a JDK in Fedora these days? ...
17:10:12 <sgallagh> I don't think that part is really up for debate
17:10:18 <decathorpe> :D
17:10:22 <zbyszek> Yes, we definitely want jdk in Fedora.
17:10:25 <sgallagh> Much as I'm not a fan, Java is critical to the ecosystem
17:10:25 <mhroncok> let's put JDK to flatpak!
17:10:37 <zbyszek> I also think that we want it to be "Java".
17:10:37 <jvanek> I like dnf update java more then wget random tarball  even if content is nearly the same
17:10:40 <nirik> So my big initial objection was more releng work, but thats solved with the sidetag dance.
17:11:09 <zbyszek> There are clearly some people who don't care, but there were also voices on the mailing list that some people care.
17:11:16 <jvanek> nirik, dont say twice, I may be calling for rcm help few timesmunillit soak in
17:11:32 <zbyszek> From previous expriences with firefox, I think it causes a lot of user confusion if the name is different.
17:11:37 <sgallagh> zbyszek: I *want* it to be Java, but I'm not sure I want that more than I want it to fit in with the distribution fully
17:11:51 <mhroncok> it already doesn't quite fit
17:11:59 <mhroncok> it bundles the universe
17:12:01 <nirik> well, sure, there may be some in setup or whatever, but ideally it can all be self managed...
17:12:22 <zbyszek> Yeah, I'd say it fits as well as it can, considering how Java is designed and managed upstream.
17:12:27 <mhroncok> this change is controversial in nature and naturaly folks are against it, but in reality, it does not make it any worse than it already is
17:12:46 <jvanek> sgallagh,  you can not simply workaround the certificsation
17:12:51 <nirik> well, it makes it better in the sense there's less certification/testing
17:12:58 <mhroncok> right
17:13:00 <sgallagh> mhroncok: Maybe in aggregate, but there are definitely problems being introduced.
17:13:00 <jvanek> if you ship it, youmust prove you certify the binary
17:13:10 <sgallagh> Like not honoring hardening flags when they're introduced.
17:13:12 <mhroncok> Stephen Gallagher: such as?
17:13:23 <mhroncok> any package that FTBFS is not honoring that
17:13:29 <mhroncok> until it's fixed
17:13:39 <jvanek> we already skip most of the flags as jdk od not build with  them :(
17:13:41 <mhroncok> and we let package be FTBFS for a logn time
17:13:44 <sgallagh> And we correctly recognize that as a problem case, not an acceptable situation
17:13:44 <decathorpe> (FTBFS packages are being removed after some time though)
17:14:12 <mhroncok> the time frame is "once the shipped build was built on an EOL fedora version"
17:14:12 <tstellar> jvanek: Do I understand correctly that the java-*portable packages would only exist in side-tags and never end up in the official tags?
17:14:40 <mhroncok> and I suppose Java would always be built on a supported Fedora
17:14:56 <mhroncok> (or not?)
17:14:57 <jvanek> tstellar,  I hope thy will be in normal repos
17:15:03 <jvanek> so the srpm rebuild of whole jdk will remain possible
17:15:18 <jvanek> tstellar, right now, they are in normla repos
17:15:50 <mhroncok> (brb)
17:15:51 <tstellar> jvanek: Ok, reading https://fedoraproject.org/wiki/Changes/BuildJdkOncePackEverywhere#theoretical_tagging_solution seems to me like they never get tagged to the official tags.
17:16:17 <sgallagh> mhroncok: I see what you're saying. I do. But at the same time, I'm pretty sure I wouldn't be satisfied if any core piece of software (glib2, maybe?) was FTBFS for more than a couple months
17:16:26 <jvanek> mhroncok,  you are spiritual father ofthat tagging
17:16:42 <nirik> They would still be shipped, but only in the oldest release I think?
17:16:45 <zbyszek> So I'm in general +1 to the proposal. With the side tag dance, the end result to the user is almost like status quo, except that the rebuild was done in an older Fedora. But we get proper dist tags and normal updates and bodhi testing.
17:16:50 <jvanek> mhroncok,  will the portables be ble to make it to norm repos?
17:17:09 <jvanek> I read it as yes, but not prooved it
17:17:37 <zbyszek> "6. untag the result of (2) from all the side tags from (1)"
17:17:55 <jvanek> i see
17:17:56 <zbyszek> This sounds like the builds would never become available in those newer releases.
17:18:01 <jvanek> hmm
17:18:05 <sgallagh> I guess I'm willing to live with this approach, if only because we have to live in the real world and not a good one.
17:18:06 <nirik> bodhi will not let you attach 1 build to multiple updates.
17:18:20 <jvanek> then 6 should be reworked to be ble to land in updates
17:18:56 <nirik> thats back to manual releng and bypassing the entire updates pipeline tho. :(
17:18:59 <mhroncok> (rbr)
17:19:05 <zbyszek> Hmm, but why do we need this? Shouldn't the rebuilds from 5. be enough?
17:19:06 <sgallagh> nirik: The logic must exist, since Modules did so
17:19:14 <sgallagh> Maybe that's reusable somehow?
17:19:35 <jvanek> nirik,  I really do not want to bypss testing pipeline
17:19:41 <nirik> it's possible bodhi could be modified, but in general we don't want to allow that case.
17:19:45 <jvanek> it is saving to much bugs
17:19:48 <nirik> nor do I.
17:20:11 <mhroncok> yeah, if we want to ship the thing everywhere, manual releng work is needed
17:20:14 <sgallagh> nirik: It would actually be a fairly useful feature beyond this case.
17:20:29 <nirik> how? it could really break things.
17:20:31 <sgallagh> Consider things like shim where we have no choice but to ship the same build in multiple releases.
17:20:35 <mhroncok> Stephen Gallagher: I once retitred gettext becasue it was FTBFS for 2 years
17:20:41 <mhroncok> in reality, nobody cares
17:20:47 <mhroncok> until it's gone
17:21:18 <zbyszek> sgallagh: nah, that's just how the package build was set up. There are better ways to do it.
17:21:39 <mhroncok> being able to bodhi one build into multiple Fedoras would be cool for both this and shim
17:21:39 <nirik> anyhow, how about a placeholder package/build in the newer releases that has a README explaining the flow and directing people to the package in the oldest supported release?
17:21:53 <nirik> it is very alien to bodhi's workflow...
17:21:57 <mhroncok> but I can imagine it's wuite impossible
17:22:14 <zbyszek> nirik: or just put the README in the package repo so that it shows up on src.fp.o? I think that's enough.
17:23:13 * decathorpe notes that we have spent 15 minutes on this topic
17:23:13 <tstellar> I think to avoid manual intervention, you would need to repack the *-portable build from the oldest release into *-portable builds for the later releases.  Then you would have *-portable packages for every release that you could ship.  So that would mean re-packing twice.
17:23:22 <nirik> allowing it would mean we would get cases where people attached build from release X to release Y and broke a bunch of things.
17:23:43 <jvanek> ugh
17:23:46 <nirik> I could see it's use in some exceptional cases (this one, shim)
17:23:56 <sgallagh> nirik: To be fair, plenty of people attach trivial rebuilds from release X to release Y that has the same effect.
17:24:12 <sgallagh> (I'm as guilty as anyone of that)
17:24:36 <sgallagh> Poor choice of word
17:24:46 <nirik> lots of those fail at build time tho if things are not right.
17:25:00 <sgallagh> I think we're on a tangent though
17:25:07 <nirik> yeah.
17:25:51 <nirik> I'm a weak +1 to this... and would be ok sorting out placeholder package or getting bodhi to handle the case down the road or just not shipping it in newer releases.
17:26:08 <sgallagh> If we were to give only the Java SIG tag permission for the OpenJDK package and rely on them (not releng) to manage the appropriate tagging, would alleviate the "extra work" concern?
17:26:37 <sgallagh> * tagging, would that alleviate the
17:26:48 <nirik> There's no partial permissions... The java SIG would have to be koji admins, able to tag and untag anything.
17:27:06 <mhroncok> admin is not enoguh
17:27:13 <nirik> I mean we could make a tag permission for the base tags, but that would break bodhi/other thigns.
17:27:17 <nirik> and also signing...
17:27:19 <mhroncok> I am admin and I cannot (un)tag things
17:27:31 <nirik> you can. :) just use --force :)
17:27:35 <sgallagh> How do we handle things like the grub2 tagging?
17:27:45 <mhroncok> I can ship updates from SRPMs, but not tag exisitng builds :D
17:27:45 <tstellar> I think the build once tag everywhere build process is something other maintainers would like to do too, so it seems to me like we really should have official support for this in bodhi.
17:27:49 <sgallagh> Because I can do that, but not most other tagging events that I know of
17:28:02 <nirik> it's a seperate perm, we could do that here, but it has to be the base tag... ie, f38, and f38-updates.
17:28:26 <nirik> I'm not in favor of more manual work for anyone.
17:28:28 <mhroncok> can we vote ont he general idea?
17:28:30 <sgallagh> ok
17:28:33 <mhroncok> s/ont/on/, s/he/the/
17:28:35 <zbyszek> Yes, please.
17:28:38 <sgallagh> Just spitballing.
17:28:57 <decathorpe> concrete proposal please :)
17:28:59 <sgallagh> mhroncok: Mind phrasing a proposal?
17:29:05 <mhroncok> let me...
17:29:18 <zbyszek> proposal: The change is approved. Jvanek will work with rel-eng on figuring out the best way to handle tagging and update generation.
17:29:44 <jvanek> that escalated quickly
17:29:45 <jvanek> oook
17:29:56 <mhroncok> PROPOSAL: The change is approved by FESCo, assuming releng folks and change authors figure out a way to do this that works for both parties
17:30:09 <mhroncok> sorry zbyszek, was typing
17:30:19 <decathorpe> (let's just assume these two are equivalent)
17:30:23 <zbyszek> I guess both things mean the same… Yeah.
17:30:26 <mhroncok> the spirit of both proposal is identical
17:30:39 <mhroncok> s/proposal/proposals/
17:30:42 <zbyszek> I'll withdraw mine.
17:30:45 <zbyszek> mhroncok: +1
17:31:00 <sgallagh> +1. I don't like it, but I can't come up with a better, acceptable alternative
17:31:03 <mhroncok> +1 to ym proposal
17:31:06 <decathorpe> -1
17:31:14 <mhroncok> * +1 to me proposal
17:31:17 <mhroncok> * +1 to my proposal
17:31:21 <nirik> Yeah, +1, not a nice world we live in, but we have to do the best we can with what we have.
17:31:25 <zbyszek> (+1 to yum proposal)
17:31:33 <mhroncok> (the matrix edits are probably horrible on IRC, sorry)
17:31:49 <zbyszek> Nah, they just appear as '* <edited form>'
17:31:57 <tstellar> I'm -1 on this.  Partly because I'm new and haven't had a lot of time to think about it, but I also I think the build/tagging flow needs to be more better defined.
17:32:19 <mhroncok> tstellar: does it need to be defined by us?
17:33:06 <jvanek> nirik, https://pagure.io/releng/issue/11438 will be source of our discussion?
17:33:15 <sgallagh> Unless I miss my guess, it's not possible for this to pass today, given only a six-person quorum and two -1 votes.
17:33:31 <nirik> jvanek: sure, it can be...
17:33:33 <tstellar> mhroncok: I think it needs to be more clear in the proposal how the process will work.  But also for me, a better defined process won't make this an automatic +1, it will just help me feel like I understand the proposal better.
17:33:52 <decathorpe> tstellar: you're remaining at -1 for now?
17:33:59 <tstellar> decathorpe: Yes, I'm at -1.
17:34:12 <jvanek> tstellar,  if you lack understanding, you may stay on 0?
17:34:22 <mhroncok> it doesn't matter
17:34:25 <zbyszek> This doesn't matter. 0 is the same as -1.
17:34:29 <tstellar> jvanek: -1 and 0 are effectively the same.
17:34:34 <jvanek> oook
17:34:50 <mhroncok> what are the next steps?
17:34:52 <nirik> so, here's a example of something that didn't pass or get rejected... so, can we discuss the releng / tagging side and rediscuss next week?
17:34:53 <decathorpe> side note: can we get rid of that 0 vote? if it's redundant? it's just confusing ...
17:35:09 <zbyszek> decathorpe: let's discuss this at some other time.
17:35:12 <decathorpe> right now we are at: (+4, 0, -2)
17:35:12 <jvanek> nirik,  I would say so
17:35:16 <tstellar> jvanek: One thing I would recommend is take to bohdi mantainers about possible changes to make ths process easier.
17:35:20 <mhroncok> Fabio Valentini: I vote 0 to your proposal!
17:35:23 <tstellar> s/take/talk/
17:35:53 <decathorpe> anybody got the voting rule book nearby?
17:36:00 <mhroncok> yes
17:36:00 <nirik> jvanek: I can try and add some options to the ticket for you / your team to discuss...
17:36:01 <tstellar> jvanek: I know that's a longer term play, but I think the idea of build once tag anywhere comes up enough that it would be worth getting first class support for it.
17:36:05 <mhroncok> my proposal is REJECTED
17:36:25 <jvanek> nirik, tstellar yy:(
17:36:26 <mhroncok> I have a counterproposal
17:36:50 <decathorpe> #agree REJECTED (+4, 0, -2): Change proposal is not accepted as-is
17:36:57 <mhroncok> the change owners will discuss the actual workflow with rleng and update the change proposal accordingly, resubmitting it when ready
17:37:11 <nirik> I think that might be nice, but I am not sure how much bodhi work is involved or if anyone can commit to it.
17:37:25 <decathorpe> mhroncok: is that a separate proposal we can vote on? ;)
17:37:32 <mhroncok> yes
17:37:33 <zbyszek> mhroncok: +1
17:37:33 <jvanek> nirik, tstellar jednorozec oook... see you  at https://pagure.io/releng/issue/11438 then
17:37:37 <decathorpe> +1
17:37:42 <mhroncok> +1
17:37:56 <nirik> uh, sure, +1
17:39:01 <tstellar> +1 and I'd also like for the change owners to get feedback from bodhi developers about feasibility of supporting this in bodhi.  Even if the answer is "we can't do this"  I think that helps the proposal.
17:39:59 <decathorpe> :thumbsup:
17:40:07 <decathorpe> sgallagh: your opinion?
17:40:37 <sgallagh> Sorry, got nerd-sniped by the 0 topic :-)
17:41:11 <sgallagh> That isn't really a proposal, it's just a deferral of the vote. But sure, +1
17:41:18 <decathorpe> thanks
17:41:33 <mhroncok> it is not a deferral of the vote
17:42:02 <mhroncok> when i said resubmit, I actually meant the whole shebang
17:42:22 <nirik> you mean the entire change process?
17:42:29 <mhroncok> (I asked myself: what would Ben want)
17:42:35 <mhroncok> yes
17:42:47 <mhroncok> sorry fi that was not clear
17:42:50 <mhroncok> *if
17:43:17 <nirik> I think thats overkill and I am not sure I understand how it was rejected by not passing... but I'll shutup and go with the flow. ;)
17:43:56 <mhroncok> (happy to be corrected, but I assumed a proposal that does not gain +5 is REJECTED)
17:44:01 <decathorpe> (I think the "resubmitting it when ready" in mhroncok's proposal was pretty clear wrt/ the intentions)
17:44:03 <sgallagh> nirik: See my latest attempt to clarify the voting rules over in #meeting:fedoraproject.org 😛
17:44:16 <nirik> yeah, it's a muddle for sure.
17:44:17 <mhroncok> we are in Fedora Metting
17:44:20 <sgallagh> err, #devel:fedoraproject.org
17:44:26 <sgallagh> Sorry, post-lunch brain
17:44:27 <mhroncok> uh, Meeting
17:44:37 <decathorpe> #agree AGREED (+6, 0, -0) The change owners will discuss the actual workflow with rleng and update the change proposal accordingly, resubmitting it when ready.
17:44:50 <decathorpe> any objections to this statement? I can still UNDO
17:45:23 <mhroncok> no objections except my own typos
17:45:44 <decathorpe> I kept those to make sure I don't mess things up ;)
17:46:03 <sgallagh> FESCo is bug-for-bug compatible!
17:46:08 <mhroncok> :)
17:46:31 <nirik> as long as jvanek is clear on it. ;)
17:46:33 <mhroncok> (if you folks rather punt the vote and let us revote next week, that's also an option)
17:46:41 <jvanek> yy am reading it
17:46:45 <decathorpe> no backsies!
17:46:49 <mhroncok> :D
17:47:00 <decathorpe> This is an orderly establishment!
17:47:15 * sgallagh quickly hides the silverware behind his back
17:47:19 <decathorpe> </British accent>
17:47:34 <decathorpe> ok, let's move on, I think it's enough for today
17:47:44 <decathorpe> #topic Next Week's Chair 🪑
17:47:44 <mhroncok> indeed
17:47:44 <jvanek> yy../me dead
17:48:09 <mhroncok> no fesco'ing for me next week, I have a PTO
17:48:23 <zbyszek> I'll be travelling next week, so I might miss the meeting.
17:49:15 <decathorpe> that might mean we don't have quorum next week anyway?
17:50:29 <sgallagh> Next week is also a US holiday
17:50:41 <sgallagh> However we have one more order of business to cover today.
17:50:54 <sgallagh> Since we have a change in membership, we have to identify whether this meeting time is still acceptable
17:51:32 <zbyszek> Not "acceptable". We have to figure out if this meeting time is the least bad ;)
17:51:38 <decathorpe> Proposal: We skip next week, and determine whether this meeting time is still acceptable, and hold the meeting at the new? time in two weeks.
17:51:48 <sgallagh> +1
17:51:48 <zbyszek> decathorpe: +1
17:52:22 <nirik> sure, works for me.
17:52:25 <decathorpe> Any volunteers for setting up a whenisgood poll / running Next(Next Week's Meeting)?
17:52:41 <zbyszek> I'll do it.
17:52:51 <decathorpe> great, thanks
17:53:11 <decathorpe> #action zbyszek to set up a whenisgood poll and chair FESCo meeting on June 27
17:53:21 <decathorpe> #topic Open Floor
17:53:23 <zbyszek> No, wait.
17:53:30 <mhroncok> is it gonna be on June 27?
17:53:31 <zbyszek> I can't chair on the 27th.
17:53:41 <decathorpe> well 13 + 14 = 27 ...
17:53:55 <decathorpe> assuming it's still Tuesday
17:54:06 <sgallagh> I can if that's the selected day
17:54:17 <sgallagh> But we should probably defer on selecting a chair until we know the day and time
17:54:19 <zbyszek> Sorry, I can do the poll, but I can't commit to anything for the next three weeks.
17:54:30 <decathorpe> #undo
17:54:30 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x7f1e017d12e8>
17:54:58 <decathorpe> #undo!
17:54:58 <zodbot> Removing item from minutes: ACTION by decathorpe at 17:53:11 : zbyszek to set up a whenisgood poll and chair FESCo meeting on June 27
17:55:17 <decathorpe> :D
17:55:28 <decathorpe> #action zbyszek to set up a whenisgood poll
17:55:48 <decathorpe> #info Chair for the meeting in ~two weeks will be determined based on the new meeting day / time
17:55:54 <decathorpe> #topic Open Floor
17:56:18 <sgallagh> Do we want to discuss the voting rules formally, or shall I open yet another ticket? :)
17:56:41 <decathorpe> not sure if I should answer that
17:56:44 <zbyszek> I think a ticket would be better.
17:56:53 <nirik> +1
17:56:56 <zbyszek> If we start discussing rules, it'll be another hour here.
17:56:58 <nirik> 😀
17:57:05 <decathorpe> better a ticket than to make this meeting even longer
17:57:19 <sgallagh> Will do, then
17:57:27 <decathorpe> (why do I always end up running the Java / Modularity / <dumpster fire of the day> meetings??)
17:57:30 <mhroncok> 0 to discussing the voting rules here
17:57:30 <nirik> Or a PR?
17:58:05 <decathorpe> anyway I think we're done for today. I will endmeeting unless there are objections within the next few seconds ;>
17:58:32 <sgallagh> Fabio Valentini: Just lucky, I guess
17:59:01 <decathorpe> #endmeeting