15:01:06 <nils> #startmeeting modularity_wg
15:01:06 <zodbot> Meeting started Tue Nov 28 15:01:06 2017 UTC.  The chair is nils. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:06 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:01:06 <zodbot> The meeting name has been set to 'modularity_wg'
15:01:07 <nils> #meetingtopic Meeting of the Modularity Working Group (once every two weeks)
15:01:07 <nils> #chair dgilmore langdon mikedep333
15:01:07 <zodbot> Current chairs: dgilmore langdon mikedep333 nils
15:01:14 <langdon> .hello2
15:01:18 <nils> #topic Roll Call
15:01:24 <nils> .hello nphilipp
15:01:25 <sct> .hello sct
15:01:27 <zodbot> langdon: langdon 'Langdon White' <langdon@redhat.com>
15:01:30 <zodbot> nils: nphilipp 'Nils Philippsen' <nphilipp@redhat.com>
15:01:30 <nils> #chair sct
15:01:30 <zodbot> Current chairs: dgilmore langdon mikedep333 nils sct
15:01:33 <zodbot> sct: sct 'Stephen Tweedie' <sct@redhat.com>
15:02:08 <asamalik> .hello2
15:02:09 <zodbot> asamalik: asamalik 'Adam Samalik' <asamalik@redhat.com>
15:02:36 <sgallagh> .hello2
15:02:37 <nils> #topic Agenda
15:02:37 <nils> #info [langdon]: Stream naming convention updates?
15:02:37 <nils> #info [langdon]: Thoughts on modular server release
15:02:38 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
15:03:09 <nils> these two are (still) open issues on Pagure, anything else?
15:03:33 <nils> ok
15:03:42 <nils> #topic Stream naming convention updates?
15:03:55 <langdon> nils: i would re order those, no?
15:04:02 <asamalik> +1 langdon
15:04:04 <langdon> like i think thte second is higher priority
15:04:04 <nils> sure :(
15:04:07 <nils> :) even
15:04:09 <nils> #undo
15:04:09 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x50e41d90>
15:04:11 <langdon> ha
15:04:13 <nils> #undo
15:04:13 <zodbot> Removing item from minutes: INFO by nils at 15:02:37 : [langdon]: Thoughts on modular server release
15:04:17 <nils> #undo
15:04:17 <zodbot> Removing item from minutes: INFO by nils at 15:02:37 : [langdon]: Stream naming convention updates?
15:04:24 <nils> #info [langdon]: Thoughts on modular server release
15:04:30 <nils> #info [langdon]: Stream naming convention updates?
15:04:36 <nils> #topic Stream naming convention updates?
15:04:52 <nils> so, that's better
15:04:57 <langdon> so.. discussion is happening here: https://pagure.io/Fedora-Council/tickets/issue/149
15:05:55 <langdon> basically, concerns have been raised that the incompleteness of content and a low level re-architecture should delay/supercede/remove the f27 server release
15:06:06 <langdon> and instead do a child-of-boltron release
15:06:35 <langdon> and maybe ship regular f27 server from the builds we did on the traditional side
15:07:12 <langdon> discuss!
15:07:23 <asamalik> yeah... and also a proposal of a new approach to modularize F28: https://asamalik.fedorapeople.org/modularity-hybrid.png
15:07:46 <langdon> asamalik: or f27 :)
15:08:31 <asamalik> langdon: ha.. right.. :) I should say ...to modularize F28 which leads us to a Q: should we also do the same thing for F27? :)
15:08:40 <langdon> i am not sure we have many people around :(
15:09:41 <asamalik> yeah.. :(
15:09:55 <nils> langdon, people who haven't discussed this in the ticket will have to catch up what was discussed already
15:10:04 <asamalik> ha
15:10:09 <nils> lest we want to rehash this here :)
15:11:18 <langdon> Not sure.. Maybe the ticket is enough.. I just wanted to be sure this group saw the discussion
15:11:30 * tflink is still reading the ticket
15:11:48 <asamalik> probably... also sgallagh wants to talk about this in the Server WG meeting later today
15:11:53 <sgallagh> Yes
15:11:55 <nils> And I'd like to forestall any forays in the field of "but we can do this already with -compat packages" which seems to happen in the ticket :D
15:11:59 <sgallagh> Sorry, was distracted. I am here now
15:12:08 * asamalik welcomes sgallagh
15:12:36 <sgallagh> The crux of the issue is that I have reservations about shipping what we have under the name Fedora Server Edition
15:13:09 <sgallagh> In large part because I suspect that its lack of access to much of the endpoint software in Fedora is going to be the thing that all press focuses on
15:13:50 <sgallagh> Combined with the fact that we're reworking the plan in F28, I think it might be more prudent to acknowledge publicly that we tried something and it didn't pan out
15:14:19 <sgallagh> Rather than release something with the implied support(*) of Server Edition.
15:14:30 <nils> mhm
15:15:12 <sgallagh> My perspective is that instead of shipping Fedora 27 Server Edition GA, we should switch our efforts immediately to Fedora 28 Server Edition and do a series of Alpha releases, converging again on the F28 Beta schedule
15:15:52 <asamalik> sgallagh: that makes sense to me... it of course doesn't mean we don't want to release it at all... maybe under a different name... but admitting it and making sure F28 will work out well is something that sounds right to me
15:15:53 <nils> sgallagh, switch all modules which are "f27" today to "f28" now and work from there?
15:16:16 <langdon> The big tradeoffs on that are 1) can't test updates through bodhi 2) smaller amount of feedback
15:16:18 <sgallagh> nils: not *exactly*
15:16:32 <sgallagh> langdon: Why can't we test updates through Bodhi?
15:16:36 <sgallagh> We can do that *right now*?
15:16:42 <sgallagh> err, s/?/!/
15:17:06 <langdon> What would we update?
15:17:12 <sgallagh> I've verified that modules submitted through Bodhi end up in updates-testing
15:17:19 <sgallagh> I proved that out during the beta freeze
15:17:38 <sgallagh> langdon: While we are in Freeze, it's the best way to try out new changes before stuffing them into the compose
15:17:41 <langdon> Sure.. But that isn't at scale.. Just negative testing
15:18:10 <langdon> Maybe it is enough.. But it makes me nervous
15:18:32 <sgallagh> nils: Switch app-stream modules to depend on platform: f28 at least.
15:18:46 <sgallagh> We're going to be changing (significantly) what platform:f28 means
15:18:55 <sgallagh> It'll basically become the same as F28 traditional
15:19:04 <asamalik> nils: also, making them smaller, removing all the bundled dependencies, because they will be in the much bigger "base" thing
15:19:04 <sgallagh> Which also means: no more bootstrap module
15:19:37 <nils> sgallagh, because that'll be gobbled up by "son of platform"?
15:19:39 <asamalik> nils: so no more dependency hell to deal with
15:19:46 <sgallagh> Well, the technical implementation is still a bit TBD.
15:19:54 * sgallagh just had a few thoughts that I'll save for post-meeting
15:19:55 <nils> asamalik, that sounds sweet
15:21:17 <sgallagh> nils: More or less; I just had a few ideas around "son of platform" that we need to work out.
15:21:33 <asamalik> on the other hand, and this is just a crazy idea to save F27.. could we generate a module that would exist on top of existing platform, containing all the dependencies for the leaf modules?
15:21:45 <langdon> How much would the decision to shift focus to f28 be obviated if we had stream expansion?
15:21:50 <sgallagh> It might be that we replace "bootstrap" with "everything" and then "platform" is a limited subset of "everything" that we know we can swap out sometimes.
15:21:54 <sgallagh> But that's a rat-hole
15:22:36 <sgallagh> asamalik: We *could*, but it's harder to implement than it sounds.
15:22:38 <asamalik> langdon: it wouldn't help at this situation, because platform f27 and f28 would be significantly different
15:22:47 <asamalik> like containing some of the deps vs. all of them...
15:23:17 <asamalik> so building the same thing on top of both wouldn't work... I think
15:23:25 <asamalik> sgallagh: yes, probably
15:23:39 <asamalik> that's why it's just a crazy idea :)
15:24:17 <sgallagh> asamalik: Well, part of the stream expansion thing is that we'd get notifications for each platform that actually failed.
15:24:30 <sgallagh> But you're right, that number would be very close to 100% at this juncture
15:25:58 <asamalik> yeah.. the way I see it, stream expansion is very useful for building against two versions of the same thing... not two completely different things
15:28:08 <asamalik> so yeah... if we could do an "environment with deps" in f27, then we could build the "app" modules against that and reuse them in F28 as well...
15:28:31 <asamalik> and the "environment with deps" would be basically a huge platform, or a shared-userspace module, or something similar
15:28:50 <asamalik> but it would be an extra work with implementing the same a different way, maybe? I don't know..
15:28:58 <sgallagh> asamalik: In other words... exactly the thing we plan to do in F28...
15:29:07 <asamalik> we would need a few days to play with it I guess... and to think
15:29:23 <sgallagh> Which is a significant undertaking, which I think is better spent getting right for F28, rather than trying to shoehorn into F27 so late
15:29:42 <asamalik> sgallagh: right...
15:29:52 <asamalik> yeah :(
15:32:59 <langdon> So.. Share your thoughts in the ticket and we move on?
15:33:24 <asamalik> unrelated... if we decided to go for F28, could we target releasing it earlier than other editions? would that look better?
15:33:46 <asamalik> langdon: yeah, probably.. we're not deciding here anyway
15:33:55 <tflink> I'd think that a good release would be good enough, no? would it really need to be early?
15:33:59 <sgallagh> asamalik: I'd prefer not to make scheduling decisisions like that
15:34:11 <tflink> even better point
15:34:34 <asamalik> that was just another crazy idea, agree
15:35:19 <asamalik> ok, so moving to the next topic?
15:35:28 <asamalik> considering we leave our comments in the ticket?
15:35:52 <nils> langdon, so #link the ticket and #info that we'll leave comments there?
15:36:44 <langdon> Nils could you? I'm on mobile for a few
15:36:50 <nils> sure
15:37:25 <nils> #link https://pagure.io/Fedora-Council/tickets/issue/149 Discussion is happening here
15:37:46 <nils> #info We'll continue discussion in the ticket
15:37:46 <langdon> Thanks
15:37:54 <nils> alright
15:38:13 <nils> #topic Stream naming convention updates?
15:38:45 <nils> https://pagure.io/modularity/issue/75
15:39:11 <nils> langdon, I just listed that because the ticket is still open and I'm not sure if we updated the Module Packaging Guidelines yet
15:39:25 <langdon> No.. Not yet :(
15:39:45 <nils> ok , I'll leave it open then
15:39:53 <langdon> I think i sent them out though? Right?
15:40:09 <nils> IIRC you sent something out
15:40:27 <asamalik> yeah it looks pretty good, there are just few comments that need to be sorted out
15:40:33 <asamalik> mostly minor changes
15:40:41 <langdon> K I'll try to follow up today
15:41:30 <nils> langdon, I'll try to remember closing the ticket when you're done, deal?
15:41:48 <langdon> Ha.. Deal
15:42:00 <nils> langdon, let me know if there's anything I can help you with
15:42:09 <nils> ok...
15:42:15 <nils> #topic Open Floor
15:42:23 <nils> anybody got anything else to discuss?
15:44:26 <asamalik> maybe come to the Server WG later today? just a reminder :)
15:44:30 <asamalik> but I don't have anyting else
15:44:53 <nils> yeah, seems y'all can have back a quarter of an hour :)
15:44:55 <langdon> I'll be there
15:44:59 <nils> #endmeeting