16:02:29 <dcavalca> #startmeeting Fedora ELN SIG (2021-06-04)
16:02:29 <zodbot> Meeting started Fri Jun  4 16:02:29 2021 UTC.
16:02:29 <zodbot> This meeting is logged and archived in a public location.
16:02:29 <zodbot> The chair is dcavalca. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:29 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:02:29 <zodbot> The meeting name has been set to 'fedora_eln_sig_(2021-06-04)'
16:02:35 <cyberpear> .hi
16:02:36 <zodbot> cyberpear: cyberpear 'James Cassell' <fedoraproject@cyberpear.com>
16:02:41 <dcavalca> #meetingname eln
16:02:41 <zodbot> The meeting name has been set to 'eln'
16:02:50 <tdawson> .hi
16:02:51 <zodbot> tdawson: tdawson 'None' <tdawson@redhat.com>
16:02:59 <michel> .hello salimma
16:03:00 <zodbot> michel: salimma 'Michel Alexandre Salim' <michel@michel-slm.name>
16:03:00 <dcavalca> #chair sgallagh bookwar[m] dcavalca
16:03:00 <zodbot> Current chairs: bookwar[m] dcavalca sgallagh
16:03:10 <dcavalca> #topic init process
16:03:51 <dcavalca> running the show on behalf of sgallagh today who's dealing with a personal emergency
16:03:59 <dcavalca> lemme find the agenda
16:05:22 <dcavalca> 1. Broken Compose and detection (sgallagh)
16:05:23 <dcavalca> 2. Tracking side-tags in Rawhide (sgallagh)
16:05:23 <dcavalca> 3. ELN branch request policy (dcavalca)
16:05:23 <dcavalca> 4. Clarifying guidance on conditionals (ngompa)
16:05:53 <dcavalca> but first
16:05:55 <dcavalca> #topic Old Business
16:06:06 <dcavalca> anybody has updates on what we discussed the last time?
16:06:33 <tdawson> Not I
16:07:23 <dcavalca> the only update on my side is that I put together a draft for the branch request policy, but we can discuss that once we get to the agenda item
16:08:07 <dcavalca> alright, let's move to
16:08:09 <dcavalca> #topic New Business
16:08:18 <dcavalca> let's start with the first item
16:08:25 <dcavalca> #topic Broken Compose and detection
16:09:03 <dcavalca> tdawson: posted a preliminary script for this on devel
16:09:11 <tdawson> I have a simple script written - https://tdawson.fedorapeople.org/eln/eln-compose-status.py - but the heart of it is just grabbing the json from here - https://odcs.fedoraproject.org/api/1/composes/?source=eln%23eln&compose_type=production
16:09:50 <dcavalca> tdawson: this seems sensible to me, what are the next steps here?
16:10:21 <tdawson> I guess to find a place to run it, and at the very least have a web page, and then work on notifications.
16:10:41 <dcavalca> tdawson: would this be a good fit for toddlers?
16:11:05 <tdawson> Actually, I think it would, once we figure out where to put it, I'm thinking somewhere on Jenkins.
16:11:35 * dcavalca is not terribly familiar with what infra is available, but this sounds reasonable
16:11:38 <tdawson> The basics of it is just grabbing a json file, parsing it, and ... doing stuff with what you parse.
16:12:12 <jforbes> Is it giving a diff between the last compose, or just a status of the current?
16:12:24 <jforbes> .hello
16:12:24 <zodbot> jforbes: (hello <an alias, 1 argument>) -- Alias for "hellomynameis $1".
16:12:33 <jforbes> .hello2
16:12:34 <zodbot> jforbes: jforbes 'Justin M. Forbes' <jforbes@redhat.com>
16:12:44 <dcavalca> based on https://pagure.io/fedora-infra/toddlers it looks like there's already an ansible thing to set these up on infra
16:12:46 <tdawson> It's got the status of the last 10 composes
16:12:54 <dcavalca> so maybe you can just tie into that
16:12:55 <tdawson> Or I should say, last 10 compose attempts.
16:13:36 <dcavalca> tdawson: for notification, I guess fedmsg + email is probably enough?
16:13:42 <tdawson> Oooohhh ... I thought you meant was it something we could have a new person do.
16:13:52 * michel very amused by the name
16:14:10 <dcavalca> tdawson: ahah trust me, I was as confused as you are now when someone mentioned this to me the first time
16:14:46 <tdawson> Well, it could run off fedora-messaging, but I was thinking of just once a day, since composes now run once a day.
16:15:13 <bookwar[m]> .hello bookwar
16:15:17 <zodbot> bookwar[m]: bookwar 'Aleksandra Fedorova' <alpha@bookwar.info>
16:16:02 <dcavalca> tdawson: I'm told there's a way to do cron-like things with toddlers but I haven't tried it myself
16:16:08 <dcavalca> https://pagure.io/fedora-infra/ansible/blob/main/f/roles/openshift-apps/toddlers/templates/cron_playtime.yml is the example I was pointed towards
16:16:42 <bookwar[m]> On the place to run it - we can use Fedora CI Jenkins. We can send messages from there too. So consider it an option
16:16:56 <tdawson> OK, I was more leaning towards putting it in eln/scripts and then have it run on eln-periodic in jenkins.
16:17:05 <dcavalca> yeah, infra-wise I'm fine with whatever folks are comfortable with
16:17:49 * cyberpear notices zodbot didn't set the topic
16:18:12 <tdawson> dcavalca: I'll look into toddlers, and see how easy/hard it is.
16:18:43 <dcavalca> do we need a vote on this? I'm inclined to just leave it to tdawson to sort out as he's done all the work so far :)
16:18:48 <jforbes> "so easy a toddler could do it?"
16:18:51 <bookwar[m]> btw, i tried to render json produced by jenkins into a nice table on the Fedora Docs site, but apparently there is some security measure which prevents loading json from external resource there
16:19:05 <tdawson> Though, I'm still leaning towards Jenkins, since I already know how to do it, and it has a web area already.
16:19:19 <dcavalca> bookwar[m]: probably XSS protection
16:19:38 <dcavalca> there should be a way to add the Jenkins endpoint as a trusted source
16:19:51 <bookwar[m]> yeah, so we stuck with the simple iframe approach for now https://docs.fedoraproject.org/en-US/eln/status/
16:20:16 <tdawson> dcavalca: I'm ok doing a vote, but not this week.  Let me look into it, and if we want a vote, I'll show what I have by then.
16:20:35 <dcavalca> tdawson: sounds good
16:20:43 <dcavalca> I think that's all for this topic, let's move to the next one
16:20:54 <dcavalca> #topic Tracking side-tags in Rawhide
16:21:43 <Eighth_Doctor> .hello ngompa
16:21:44 <zodbot> Eighth_Doctor: ngompa 'Neal Gompa' <ngompa13@gmail.com>
16:22:10 <dcavalca> does anybody have anything on this topic?
16:22:44 * dcavalca is trying to find some context
16:23:08 <bookwar[m]> I guess it is our favorite https://github.com/fedora-eln/eln/issues/2
16:23:14 <jforbes> Well, I know they are doing the python-3.10 sidetag now, but do we have a tracking issue there?
16:23:21 <dcavalca> oh, that one
16:23:22 <dcavalca> thanks bookwar[m]
16:23:44 <bookwar[m]> but jforbes is talking about a different one?
16:24:01 <jforbes> bookwar[m]: I was just making a guess, I haven't seen any issue there
16:24:20 <bookwar[m]> ok, so the one i mentioned is about dynamic sidetags in rawhide
16:25:16 <bookwar[m]> I have a script to try there, unfortunately i got stuck on the trigger part in Jenkins, i wasn't able to subcribe to the right topic to read messages about new sidetags created
16:25:30 <bookwar[m]> I should resurrect that effort
16:25:46 <dcavalca> sounds like a plan
16:26:25 <dcavalca> anything else folks want to discuss on this?
16:26:38 <tdawson> Just that I'm all for it
16:26:39 <bookwar[m]> If we also want to support tracking of a more long-term sidetags, which are not in the form f35-build-side-37359 we should have a separate issue for that
16:27:53 <tdawson> Do those happen often?
16:28:13 <dcavalca> I supposed those could help if we wanna do things like test out a new compiler version, which might take a while
16:28:17 <jforbes> Not often, and typically very well advertised
16:28:57 <dcavalca> but yeah, agree that it's worth of a separate issue
16:29:04 <bookwar[m]> i am not sure if people use the special dedicated sidetags now, while dynamic sidetags are available to anyone. But just wanted to point out that it is a separate thing.
16:29:05 <tdawson> I'd say let's stick with getting the normal ones working good first, then go after those seperately.
16:29:13 <bookwar[m]> +1
16:29:16 <dcavalca> sounds good
16:29:26 <jforbes> +1
16:29:34 <bookwar[m]> action on me to setup the dummy pipeline, so that we can iterate on the script
16:30:04 <dcavalca> #agreed focus on tracking regular side tags first, and tackle the long-term ones later on
16:30:30 <sgallagh> .hello2
16:30:31 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
16:30:45 <dcavalca> #action bookwar[m] setup the dummy pipeline so that we can iterate on the script
16:30:54 <dcavalca> alright, I think we can move to the next one
16:31:12 <dcavalca> #topic ELN branch request policy
16:31:25 <dcavalca> as discussed in the last meeting, I put together a draft
16:31:41 <dcavalca> #info ELN branch request policy draft: https://hackmd.io/OoLbOauKR7WexZ0Rdm-XiA
16:32:11 <dcavalca> feel free to comment/edit/wordsmith as you see fit
16:32:35 <sgallagh> Thanks, dcavalca!
16:32:47 <dcavalca> my idea what to make this into a PR once we have general agreement on it
16:33:07 <dcavalca> (as it's a lot easier to do collaborative editing in hackmd)
16:33:44 <dcavalca> I'd say have a read, and we can iterate on it between now and the next meeting?
16:33:55 <dcavalca> as it's probably not worth it to use this time to bikeshed on it
16:33:57 <bookwar[m]> +1, thank you for putting it together
16:34:45 <sgallagh> Yes, that makes sense.
16:35:06 <dcavalca> alright, let's move on the next item then
16:35:10 <dcavalca> #topic Clarifying guidance on conditionals
16:35:25 <sgallagh> Eighth_Doctor: I think you had some concerns here?
16:35:40 <Eighth_Doctor> yeah
16:35:43 <bookwar[m]> Also on previous item,  we won't be able to apply the policy immediately, as we will need to do some adjustments to the automatione. So for now we are just focusing on the policy itself.
16:36:09 <dcavalca> bookwar[m]: yep, agreed
16:36:15 <Eighth_Doctor> so I had a conversation with someone yesterday when dealing with a bug caused by EL integration into Fedora
16:36:28 * sgallagh prefers to have a policy that we code the implementation against, rather than the reverse
16:36:35 <Eighth_Doctor> and that person told me that ELN SIG had said we'd recommended using %if 0%{?fedora} (indicating Fedora is the special case)
16:36:46 <Eighth_Doctor> which is something I distinctly remember us saying we don't recommend
16:37:07 <sgallagh> I don't recall us making any statements either way on that, actually
16:37:22 <dcavalca> uh, I think a bunch of folks just dropped off IRC
16:37:38 <sgallagh> We agreed that conditionals were preferable to forking an ELN branch and that the goal was to stay as close as possible to Fedora.
16:37:47 <dcavalca> sgallagh: yeah, my recollection is that we agreed that conditionals are preferred, but we didn't really specify it beyond that
16:37:47 <Eighth_Doctor> so I was wondering if we could clarify our guidance to indicate that we prefer %if 0%{?rhel} and the "normal" case is the Fedora one
16:37:50 <sgallagh> zodbot is still here, so the logs remain available
16:37:50 <Eighth_Doctor> that makes it easier for alternative Fedora downstreams to get predictably Fedora-ish behavior by default
16:37:54 <jforbes> I was the one confused on that, and for some reason I call that being discussed when ELN was created. Entirely possible that I was wrong
16:37:55 <michel> that seems like a good idea to me
16:38:19 <sgallagh> replay: We agreed that conditionals were preferable to forking an ELN branch and that the goal was to stay as close as possible to Fedora.
16:38:19 <dcavalca> Eighth_Doctor: yeah, it seems sensible to me that fedora should be default and the conditionals should be for rhel
16:38:45 <bookwar[m]> we do have a doc here https://docs.fedoraproject.org/en-US/eln/ftbfs/
16:38:50 <dcavalca> unless maybe the package is very much RHEL specific, though I'm not sure how common that is in practice
16:38:59 <sgallagh> But that being said, it probably makes sense to treat `%{fedora}` as the standard.
16:39:00 <bookwar[m]> i think we use fedora in places where we need to specify which version of fedora it applies to
16:39:17 <bookwar[m]> if we need to differentiate between old fedora and current state
16:39:41 <sgallagh> "The best spec files alter default behavior only for released versions of Fedora or RHEL, making the default policy the one applicable to the current versions of Fedora or RHEL."
16:39:41 <jforbes> Looks like rhel and fedora are considered equals in that doc, and eln is "last resort" which makes sense
16:40:01 <sgallagh> That's the closest we came to a statement on this topic, I think
16:40:52 <dcavalca> #info Proposal: The best spec files alter default behavior only for released versions of Fedora or RHEL, making the default policy the one applicable to the current versions of Fedora or RHEL.
16:40:55 <sgallagh> But as I said, I think more clarity could be better.
16:41:04 <sgallagh> That wasn't a proposal
16:41:11 <dcavalca> ah, my bad
16:41:13 <dcavalca> #undo
16:41:13 <zodbot> Removing item from minutes: INFO by dcavalca at 16:40:52 : Proposal: The best spec files alter default behavior only for released versions of Fedora or RHEL, making the default policy the one applicable to the current versions of Fedora or RHEL.
16:41:14 <sgallagh> That was the current text on the page bookwar[m] linked :)
16:41:21 <jforbes> I think the doc that bookwar[m] linked is pretty clear
16:41:32 <sgallagh> I *do* have a proposal though
16:41:41 <bookwar[m]> so current fedora should be the default, while older/previous fedora can be conditionalized?
16:41:50 <Eighth_Doctor> Yes
16:42:02 <dcavalca> bookwar[m]: yeah, that's what I'd do
16:42:08 <michel> by current, we mean Rawhide, right?
16:42:11 <michel> not current release
16:42:35 <dcavalca> good point, we should be explicit here
16:42:38 <sgallagh> Proposal: "Conditionals should be avoided where possible. The default state should be to assume the build will be performed for current Rawhide."
16:42:49 <michel> (except if the maintainer fork older branches, which is an option)
16:43:04 <sgallagh> ELN is not relevant for those branches
16:43:09 <Eighth_Doctor> sgallagh: +1
16:43:15 <cyberpear> +1
16:43:16 <dcavalca> +1
16:43:44 <Eighth_Doctor> (I assume this implies altered behavior needs to be conditioned accordingly, as I was saying earlier)
16:44:08 <tdawson> +1
16:44:44 <jforbes> Well, I don't expect anyone is using conditionals where they aren't actually needed.
16:44:53 <michel> +1
16:44:54 <tdawson> Although all we can affect is when people are considering ELN.  A Fedora packager just doing his fedora package only, can do as they please.
16:44:57 <sgallagh> jforbes: Conditionals for everyone!
16:45:31 <tdawson> That came out wrong, but I think ya'll know what I meant.
16:45:40 <sgallagh> That's a fair point as well.
16:46:00 <sgallagh> It does seem a bit as if we're making recommendations for packaging in general, which would be the purview of the FPC.
16:46:15 <tdawson> Correct, that's what I meant.
16:46:32 <jforbes> That was what I meant by I think the doc that bookwar[m] linked was already pretty clear on the matter
16:46:43 <Eighth_Doctor> well, FPC generally has the same guidance (assume default for rawhide)
16:47:04 <Eighth_Doctor> that's why I am saying this is a specific clarification for ELN
16:47:10 <Eighth_Doctor> to reinforce this
16:47:13 <dcavalca> we could always raise this with FPC to get their blessing once we've clarified our policy on the ELN side
16:47:23 * sgallagh nods
16:47:39 <dcavalca> #agreed Conditionals should be avoided where possible. The default state should be to assume the build will be performed for current Rawhide. (+6, 0, -0)
16:47:43 <tdawson> OK, if that is the FPC guidance, then I give this a double +1, just to say we want to follow their guidance.
16:47:44 <dcavalca> sgallagh: I counted you as a +1
16:48:29 <dcavalca> Eighth_Doctor: wanna take an action to submit a PR to clarify the language in the docs?
16:50:17 <jforbes> I did not vote as I live in a world of exception... ELN builds from the same SRPM and versions, but is a very different package set for me
16:51:01 <Eighth_Doctor> dcavalca: sure
16:51:39 <dcavalca> #action Eighth_Doctor submit a PR to clarify the docs as agreed
16:52:12 <dcavalca> we have 10 mins left, I guess it's time for
16:52:15 <dcavalca> #topic Open Floor
16:52:52 <tdawson> Nothing from me.  Everything I wanted to talk about, has been talked about.
16:52:56 <jforbes> Just a quick update, I had filed the issue with infra for getting ELN aws builds, it has been a while with no update, so I just pinged on the issue
16:53:22 <sgallagh> Thanks
16:54:19 <jforbes> Also, if a kernel is built for rawhide, and not built for ELN, please do not build it for me, there is probably a specific reason this was done
16:56:01 <sgallagh> jforbes: Ah, sorry about that.
16:56:20 <sgallagh> I was attempting to rebuild things that were done out of order for the Perl 5.34 mini-mass-rebuild
16:56:28 <sgallagh> I'm still having trouble there :(
16:56:59 <jforbes> Not a problem, there is just some policy around the ELN kernel which I have autonomy to work around for Fedora, but not always for ELN
16:57:20 <jforbes> typicaly not an issue, unless there is question on the direction
16:57:46 <sgallagh> (In this case, I just forgot to exclude it from the list)
16:59:45 <dcavalca> alright folks, looks like we're out of time
16:59:57 <dcavalca> have a good one everybody!
17:00:01 <dcavalca> #endmeeting