14:00:00 <nils> #startmeeting modularity_wg
14:00:00 <zodbot> Meeting started Tue May 16 14:00:00 2017 UTC.  The chair is nils. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:00 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:00:00 <zodbot> The meeting name has been set to 'modularity_wg'
14:00:00 <nils> #meetingtopic Meeting of the Modularity Working Group (once every two weeks)
14:00:00 <nils> #chair dgilmore haraldh langdon sct tflink
14:00:00 <zodbot> Current chairs: dgilmore haraldh langdon nils sct tflink
14:00:00 <nils> #topic Roll Call
14:00:07 <nils> .hello nphilipp
14:00:08 <zodbot> nils: nphilipp 'Nils Philippsen' <nphilipp@redhat.com>
14:00:11 <sct> .hello sct
14:00:12 <zodbot> sct: sct 'Stephen Tweedie' <sct@redhat.com>
14:00:54 <tflink> .hello tflink
14:00:55 <zodbot> tflink: tflink 'Tim Flink' <tflink@redhat.com>
14:01:45 <jkurik> .hello jkurik
14:01:46 <zodbot> jkurik: jkurik 'Jan Kurik' <jkurik@redhat.com>
14:02:25 <nils> #topic Agenda
14:02:32 <langdon> .hello langdon
14:02:33 <zodbot> langdon: langdon 'Langdon White' <langdon@fishjump.com>
14:02:35 <langdon> sorry im a little late
14:03:04 <nils> so, the agenda doc is pretty empty today, but langdon wanted to at least give a status update, right?
14:03:16 <langdon> nils, right
14:03:27 <nils> #info Status update
14:03:30 <langdon> basically what made it to beta freeze
14:03:33 <nils> ok, let's get at it
14:03:37 <nils> #topic Status update
14:04:09 <contyk> .hello psabata
14:04:10 <zodbot> contyk: psabata 'Petr Šabata' <psabata@redhat.com>
14:04:24 <langdon> ok.. so beta freeze was today.. and we got our list to jkaluza for inclusion
14:04:38 <langdon> we will be making the list understandable this week :)
14:04:45 <langdon> right now it is a jumble of comms
14:05:01 <langdon> adam is putting it in to a demo
14:05:26 <langdon> we also got some tickets complete with fesco
14:05:55 <langdon> so.. we are in decent shape for the release but haven't pulled together anything we can communicate effectively :)
14:06:03 <sgallagh> .hello sgallagh
14:06:04 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
14:07:36 <nils> langdon, shall I summarize for the log?
14:07:43 <langdon> yes please :)
14:07:56 <langdon> unless people have questions? sorry it is short on detailss..
14:08:11 <nils> no details is good for not getting us worried unduly :o)
14:08:18 <langdon> ha
14:08:35 <nils> alright
14:08:49 <asamalik> .hello asamalik
14:08:50 <zodbot> asamalik: asamalik 'Adam Samalik' <asamalik@redhat.com>
14:09:16 <nils> #info submitted list of modules to be included to jkaluza for beta freeze
14:09:22 * asamalik apologizes for being late... his client had better ideas than to work properly
14:09:29 <nils> #info asamalik will be putting this into a demo
14:09:45 <jkurik> langdon: thanks for informing FESCo with the status
14:10:31 <nils> jkurik, langdon: anything I should #info about that?
14:11:17 <nils> I take that as "nothing"
14:11:28 <langdon> meh.. i can list the tickets for people here though.. if you like
14:11:42 <nils> langdon, not sure how important that is
14:11:47 <langdon> right
14:11:57 <langdon> if interested, review recent fesco tickets :)
14:12:07 <nils> so... has anybody got anything they want to discuss in open floor?
14:12:25 <contyk> not sure whether "discuss" is the right word
14:12:35 <nils> contyk, ok, "something" :)
14:12:39 <nils> topic open floor
14:12:41 <nils> oops
14:12:43 <jkurik> proposed #info FESCo has been informed of the current status in https://pagure.io/fesco/issue/1706
14:12:43 <langdon> complain?
14:12:44 <contyk> but I just want to say we're still working on getting the infra-built images working
14:12:47 <nils> #topic Open Floor
14:13:08 <contyk> there's a dnf module now, there's a first dirty version of the installer module too
14:13:18 <nils> noice
14:13:20 <contyk> and we're hacking the lorax templates and kickstarts to make it all work somehow
14:13:26 <langdon> contyk, \o/ !!!
14:13:31 <langdon> sgallagh, woot!
14:13:43 <asamalik> contyk, do you have an estimate when we'll see the first dirty compose?
14:13:54 <contyk> asamalik: nope
14:13:55 <langdon> contyk, the dnf module is full dnf, right? micro is still in brt?
14:14:04 <contyk> there will be *a* compose of base-runtime:f26 (only) in prod today
14:14:04 <sgallagh> langdon: Yes
14:14:05 <nils> asamalik, "in the future" ;)
14:14:11 <contyk> and there are several dirty PoC composes in stg
14:14:32 <contyk> if you need something for development, let me or jkaluza know
14:14:37 <contyk> we can make something for you
14:14:46 <asamalik> contyk, dirty poc in stg sounds good! can we boot that somehow?
14:14:56 <contyk> asamalik: no, it's just repos
14:15:14 <contyk> langdon: yes, it's the full dnf
14:15:15 <nils> contyk, btw, we (ttomecek) noticed that there was an empty base-runtime:f26 but modules using that still built, it seemed to pull in normal F26 repos, is that known/expected?
14:15:22 <langdon> contyk, sgallagh thanks
14:15:27 <contyk> langdon: without modularity support, however; waiting until that appears in dist-git
14:15:41 <nils> or is that totally unrelated?
14:15:42 <langdon> contyk, yeah.. hmm.. that may only land in copr..
14:16:00 <contyk> nils: I don't know what you're talking about; elaborate?
14:16:28 <tflink> is there documentation around what the plans are around the more traditional artifacts is once the modularity stuff lands? stuff around how releng processes are going to change, who's responsible for the modules which will effectively define the composes etc?
14:16:59 <langdon> tflink, thats on the docket for around ga of f26
14:17:00 <tflink> there were some questions around that last week at the CI/Infra hackfest and while not an imminent issue, I was hoping that they'd at least been thought about
14:17:24 <contyk> tflink: to answer the last question, modules are registered in pkgdb co you can see the responsible people in theree
14:17:26 <tflink> langdon: all of this is supposed to land around F27, right?
14:17:26 <nils> contyk, I'm only relaying this because I overheard it on the meeting before, but ttomecek built a module of his and apparently the base-runtime he based it on was empty. I'm short on details, because that's all I know beyond that it used normal F26 packages to fulfil dependencies
14:17:28 <langdon> yeah.. been thinking about it.. working on documenting.. draft guidelines are done.. need to work through processes
14:17:41 <contyk> tflink: in extension, the component maintainers are also responsible, sort of indirectly
14:18:01 <nils> contyk, I guess if it doesn't ring a bell, I'll discuss with Tomas first and then one of us can get back to you
14:18:02 <contyk> nils: that doesn't make any sense :)
14:18:11 <langdon> tflink, i would like it to be in progress before f27.. like the changes happening between f26 & f27
14:18:13 <tflink> as far as I know, all of the releng/infra related dev cycles are already allocated for the F27 cycle
14:18:39 <tflink> if there are going to be changes requested, they really need to be communicated
14:18:53 <langdon> tflink, do you mean "software" ? or "processes"?
14:19:04 <tflink> bodhi, koji et. al
14:19:12 <langdon> factory-2 team is already working on software changes
14:19:21 <langdon> proposals, arch, code, etc
14:19:35 <langdon> the "who does what" part is more the stuff im talking about
14:20:00 <tflink> unless I misunderstood them last week, the factory-2 folks told us to ask you about these plans
14:20:07 <contyk> I think that arbitrary branching is the biggest change and that's already been filed
14:20:16 <tflink> the modularity folks in general and langdon in particular
14:20:22 <langdon> threebean, you here?
14:20:31 <contyk> https://fedoraproject.org/wiki/Changes/ArbitraryBranching
14:20:59 * threebean rejoins
14:20:59 <langdon> tflink, ill follow up with threebean and make sure we have our bases covered..
14:21:13 <contyk> they're also working on or planning to work on a few new services but that's almost entirely them
14:21:31 * tflink is really wishing he had written down all the questions and concerns that came up during that discussion
14:21:35 <contyk> e.g. freshmaker, on demand composes (does it have a name yet?)
14:21:39 <langdon> tflink, ha....
14:21:42 <contyk> ;)
14:21:42 <jkaluza> tflink: which discusion had been?
14:21:53 <jkaluza> tflink: there were lot of them last week :)
14:22:02 <jkaluza> *Which discussion was that
14:22:08 <tflink> there was a conversation around arbitrary branch support during the CI/Infra hackfest last week
14:22:15 * threebean nods
14:22:46 <langdon> tflink, that is definitely in the works.. i think factory is expecting us to define/propose the "policy" but they are the ones doing the "software changes"
14:22:47 <threebean> yeah, tflink the details there are mostly on mprahl and myself.  it should all be linked to from the ArbitraryBranching change.
14:22:52 <tflink> the conclusion that I remember was "don't ask the factory folks about the details around how arbitrary branching will affect non-modularity bits of fedora, ask the modularity folks"
14:22:53 * threebean nods
14:23:10 <threebean> langdon: correct.  we're relying on you guys to handle the policy and guidelines for how the branches should be used.
14:23:20 <tflink> langdon: have those things been defined yet?
14:23:23 <langdon> ahh i see..
14:23:39 <langdon> no.. well.. not in a real concrete way.. just architecturally
14:23:57 <langdon> thats what we want to work on ~f26-ga
14:24:06 <threebean> well - we haven't really talked about it as a group, but it makes the most sense to me to still have a 'f27' branch for all packages in Fedora for f27, but using the new arbitrary branching tooling.
14:24:30 <threebean> which would then minimize the change in workflow to the existing packager base, but would enable early adopters to experiment with "arbitrary" branches off of f27.
14:24:37 <nils> threebean, sounds fair
14:24:54 <langdon> i think we will see things slowly move over.. like.. the packages that haven't gone modular will dep on brt and shared-user-space and those branches will be called "f27" ..
14:24:58 <tflink> threebean: have the changes to bodhi been scoped out?
14:25:00 <contyk> threebean: +1
14:25:07 <threebean> tflink: just a moment, I'll link to the milestone.
14:25:33 <threebean> tflink: https://github.com/fedora-infra/bodhi/milestone/4
14:25:54 <contyk> well, regarding content, I still need to submit a change about base runtime becoming host+platform
14:26:10 <contyk> need to document that somewhere and think it through... some more...
14:26:19 <tflink> threebean: I don't see any tickets about supporting arbitrary branches during mash
14:26:34 <langdon> contyk, yeah.. and, we should probably do one for the "everything else" module.. or decide if that is synonymous with "shared userspace"
14:26:47 <threebean> hm.
14:26:55 <threebean> tflink: well, I guess it is really this one -> https://github.com/fedora-infra/bodhi/issues/1330
14:27:06 <threebean> although it was originally written to say "containers", bowlofeggs tells me that modules will come first.
14:27:13 <contyk> langdon: I expect platform will absorb most of the current shared-userspace
14:27:20 <threebean> and the arbitrary branches would actually get pulled together during module build.
14:27:29 <contyk> not sure about "everything else" module(s)
14:27:42 <tflink> I'm not trying to be a PITA for the sake of being a PITA, just concerned that there are a lot of things planned around the F27 timeframe and sometimes it sounds like the different change owners aren't always talking to eachother around requests to support services
14:27:46 <langdon> contyk, yeah.. we just need a place for "non-modularized packages" to land.. that is some how distinct to show that they should come out over time
14:27:47 <threebean> the masher then in turn would need to be modified there to pull together module builds.  bodhi would be relatively ignorant with respect to the source branches.
14:27:57 <contyk> most of the "everything else" stuff is probably in a very poor state
14:28:28 <tflink> threebean: possibly but I read that as "non-rpm", not "changing mash to support a change in how we define the contents of the distro"
14:28:33 <contyk> langdon: non-modularized packages... somehow mixed with the modularized content, you're saying?
14:28:57 <langdon> contyk, im saying "don't mix them" even though they need to be shipped
14:29:27 <langdon> sounds to me like we should do a formal "plans for now til f27" review sometime soon
14:29:32 <threebean> tflink: yeah, we talked in RDU last week about trying to replace mash in bodhi's backend with pungi.
14:29:34 <langdon> threebean, tflink ^^
14:29:39 <threebean> and if we can do that, pungi is already module-aware.
14:30:32 <tflink> threebean: and module-aware means arbitrary-branching-aware in this case?
14:30:50 <threebean> indirectly, yes.
14:31:10 <threebean> it's actually not aware of the source branches at that point.
14:31:20 <threebean> but, the point is that it shouldn't need to be.
14:31:29 <threebean> (if that doesn't make sense, I can try again)
14:31:49 <tflink> fair enough, I'm just wanting to make sure that there are plans and most folks are on the same page
14:31:56 <threebean> that's legit.
14:32:05 * tflink still remembers F18
14:32:22 <tflink> and I have absolutely no desire to go through that again
14:34:22 <tflink> threebean: for the sake of being overly thorough, there are plans with fedora releng to support creating what looks like the traditional Fedora release once modularity lands?
14:34:50 <langdon> so.. should we do a "review of the plan" session? I know i am not prepared for that at the moment but I might be able to do it in 2-3 weeks
14:35:15 <langdon> tflink, i don't think there is a "traditional release"
14:35:21 * threebean nods
14:35:29 <threebean> but the "ga" is produced by pungi.
14:35:32 <langdon> there is an "rpm only from modules" release which might feel like that
14:35:44 <tflink> langdon: whatever you call the isos and bits that people currently expect
14:35:56 <langdon> tflink, yeah.. we want to make that all from modules
14:36:22 <tflink> sure, I just want to make sure that there are plans to keep producing something that acts like what folks are used to and we're set up to test
14:36:29 <contyk> langdon: 2-3 weeks sounds good
14:36:35 <tflink> hence why I said "looks like a traditional release"
14:36:41 <threebean> tflink: yeah, that's the plan.  :)
14:36:48 <langdon> tflink, ohh yes.. even for f26 the deliverables are qcow2 and base-container-image
14:36:54 * threebean nods
14:37:12 <langdon> the repos are just the hard part (mostly) .. thats why they lead to so much discussion
14:37:21 <threebean> note that we don't have those deliverables yet, due to problems modularizing the installer.  it's being worked on, though.
14:37:30 <langdon> but sgallagh has been working on doing the installer to make qcows.. and isos are "soon"
14:37:37 * threebean nods
14:37:51 <threebean> end of the week if all goes well.  maybe not until next week, though.
14:38:13 <contyk> wonder if I should just rewrite the lorax templates from scratch
14:38:14 <threebean> "if all goes well".  that's a good one.  ;)
14:38:55 <langdon> however, tflink also raises the good point (which is in my head at least already), we need "how to qa this thing" formally in the plan
14:40:11 <contyk> langdon: automatically, duh
14:40:26 <contyk> with magic and star dust
14:40:33 <langdon> contyk, right! tflink knows all the tests can be fully automated and perfect.. im sure
14:40:57 <tflink> as I read the backscroll, I should point out that I suspect that I'm not the only one who remembers F18. that wasn't meant to make me sound special :-P
14:41:17 <tflink> langdon: in the land of unicorns and pixie dust!
14:42:14 <langdon> so.. can contyk, threebean and myself try to put the "plan" together over the next few weeks and then review here?
14:42:29 <contyk> +1
14:42:34 * langdon notes that there is a "pixie dust" box on the original f-2 arch
14:42:38 <contyk> two or four weeks from now?
14:42:43 <contyk> four sounds safer
14:42:48 <langdon> :)
14:43:23 <tflink> I'd suggest reviewing the plan with fedora releng and fedora infra sometime soon
14:43:48 <tflink> the incoming CI stuff is going to be putting stuff on their plate as well
14:44:48 <tflink> not to mention the other things that they're planning that aren't related to modularity or CI
14:45:02 <langdon> yeah.. we did that for f26.. we just haven't gotten the ducks in a row to do f27 yet
14:45:32 <tflink> FWIW, the initial scoping and planning for F27 support was done a few weeks ago
14:45:55 <tflink> which is why I'm poking at it :)
14:46:40 * langdon looks around for his hourglass to flip and get more time
14:47:11 <langdon> so.. nils anything to #info?
14:47:12 <nils> apropos, is this something that should be put on the agenda for next time, for revisiting?
14:47:21 <nils> langdon, you ask me? :o)
14:47:26 <langdon> nils, you are smart
14:48:09 <nils> langdon, actually, if I'd be smart I'd know what to #info from y'all's wall of text ;)
14:48:11 <langdon> "review plan for f27 in 4 weeks"
14:48:17 <langdon> ??
14:48:27 <nils> langdon, to #info, or agenda?
14:48:29 <langdon> i am not sure either.. threebean, contyk, tflink ?
14:49:07 <tflink> reviewing the plan and communicating it with releng and infra soon are the big ones, I think
14:49:16 * threebean nods
14:49:22 <contyk> +1
14:49:24 <langdon> ok..
14:49:49 <langdon> #info threebean, contyk, and langdon to work up a plan for f27
14:50:03 <langdon> #info plan to be reviewed by modularity-wg in 4 weeks
14:50:22 <langdon> #info plan to be reviewed with releng, infra, et al in the same timeframe
14:50:28 <langdon> how are those?
14:50:32 <nils> 👍
14:50:36 <tflink> they look good to me
14:50:38 <tflink> thanks
14:50:39 <nils> as far as I'm concerned :)
14:50:56 <nils> good, anything else?
14:51:14 <langdon> let's call it.. if we can.. only 0m left anyway
14:51:18 <nils> yeah
14:51:20 <langdon> *9m
14:51:36 <nils> so, here's the agenda doc for future meetings :)
14:51:38 <nils> https://board.net/p/modularity-wg-agendas
14:51:45 <nils> *hint*
14:51:48 <langdon> ha
14:51:52 <nils> ok, thanks everybody!
14:51:55 * asamalik waves
14:51:57 <nils> #endmeeting