15:00:09 <nils> #startmeeting modularity_wg
15:00:09 <zodbot> Meeting started Tue Aug 23 15:00:09 2016 UTC.  The chair is nils. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:09 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:09 <zodbot> The meeting name has been set to 'modularity_wg'
15:00:09 <nils> #meetingtopic Weekly meeting of Modularity WG
15:00:09 <nils> #chair dgilmore langdon tflink
15:00:09 <zodbot> Current chairs: dgilmore langdon nils tflink
15:00:24 <nils> #topic Roll Call
15:00:29 <nils> .hello nphilipp
15:00:29 <zodbot> nils: nphilipp 'Nils Philippsen' <nphilipp@redhat.com>
15:01:00 <langdon> .hello langdon
15:01:01 <zodbot> langdon: langdon 'Langdon White' <langdon@fishjump.com>
15:02:26 <cpacheco> .hello cpacheco
15:02:27 <zodbot> cpacheco: cpacheco 'Courtney Pacheco' <cpacheco@redhat.com>
15:03:30 <asamalik> .hello asamalik
15:03:30 <zodbot> asamalik: asamalik 'Adam Samalik' <asamalik@redhat.com>
15:03:41 <langdon> nils, do you have the agenda?
15:04:00 <nils> langdon: sure, once everybody's had a chance to say hello :)
15:04:14 <nils> ping dgilmore tflink sct
15:04:16 <nils> #chair sct
15:04:16 <zodbot> Current chairs: dgilmore langdon nils sct tflink
15:04:57 <sct> .hello sct
15:04:58 <zodbot> sct: sct 'Stephen Tweedie' <sct@redhat.com>
15:05:11 <nils> #topic Agenda
15:05:11 <nils> #info Ensuring we support all arches
15:05:11 <nils> #info Containers (langdon, sct)
15:05:31 <nils> anything else that wasn't on the etherpad?
15:05:47 <langdon> well.. and i am not sure we are ready to talk about containers :)
15:05:52 <nils> hah
15:05:56 <nils> #undo
15:05:56 <zodbot> Removing item from minutes: INFO by nils at 15:05:11 : Containers (langdon, sct)
15:07:01 <nils> ok, I've shelved containers for later (next week tentatively)
15:07:31 <nils> #topic Ensuring we support all arches
15:07:52 <nils> Whose is it? dgilmore, lkocman?
15:08:02 <lkocman> nils, that's mine
15:08:13 * nils cedes the floor
15:08:19 <lkocman> nils, but we don't have working rida atm to check whole chain
15:08:24 <lkocman> nils, this should change this sprint
15:08:32 <langdon> sct, more to add to the agenda?
15:08:39 <lkocman> I have a card for it ... ;-) Unless somebody merges some total rework of rida which breaks it again
15:08:45 <nils> haha
15:08:47 <langdon> i think we might have a light meeting
15:08:49 <lkocman> :-) which is very likely with new messaging
15:08:53 <sct> langdon: Not right now
15:09:18 <nils> lkocman: I mean, we can still talk about the general ideas of supporting all arches, can't we?
15:09:22 <nils> This is not a demo :).
15:10:25 <lkocman> well rida is configured to build on all that stg koji instance supports
15:10:43 <lkocman> and pungi-signed-repo reads arches from tag
15:10:48 <lkocman> so in fact we're in good shape
15:11:03 <lkocman> not sure if we need to talk more about it
15:11:03 <nils> lkocman: so I understand that as soon ride gets off the ground again, stuff should not be x86_64 only anymore
15:11:09 <lkocman> yup yup yup
15:11:15 <nils> okay, I'll #info this...
15:11:22 <lkocman> dgilmore, sounds good?
15:12:51 <nils> #info rida (the module building orchestrator) is kind of broken right now, but building for all arches (which are enabled in staging) should commence once we get it off the ground again.
15:12:58 <nils> lkocman, dgilmore: ^^
15:13:39 <nils> lkocman, dgilmore: I'll take that as a "yes" :)
15:13:55 <lkocman> nils, take it as "bravo"
15:14:01 <nils> heh :)
15:14:04 <nils> #topic Open Floor
15:14:13 <nils> anything for open floor today?
15:14:39 <lkocman> I'd mention all the documents that are supposed to involve community into modularity development!
15:14:50 <lkocman> which are on the way, and we expect great contribution
15:14:59 <lkocman> langdon, right?
15:15:06 <nils> do we have a list which we can #info?
15:15:08 <langdon> lkocman, sure?
15:15:19 <langdon> lkocman, do you mean docs? or blog posts?
15:15:21 <langdon> or both?
15:15:23 <lkocman> langdon, both
15:15:27 <lkocman> langdon, wiki + blogs
15:15:39 <lkocman> langdon, up2date component descriptions, links to projects
15:15:41 <lkocman> langdon, blog posts
15:15:44 <langdon> lkocman, yeah.. we need an announcement about the wiki content sent out
15:16:13 <lkocman> any good advice on how to reach as big audience as possible?
15:16:18 <lkocman> with these "docs"
15:16:20 <lkocman> ?
15:16:36 <cpacheco> lkocman, langdon: fedora magazine?
15:16:40 <lkocman> cpacheco, +1
15:16:41 <nils> lkocman: we should have paid for ad space at the Olympics :P
15:16:45 <cpacheco> lots of people read that magazine
15:16:56 <langdon> and not a lot of people watched the olympics
15:17:06 <langdon> well.. at least a fedora-devel mail..
15:17:18 <langdon> i had planned to do a magazine post recapping our blog posts..
15:17:29 <langdon> maybe we could add in "docs announcement" as well
15:17:45 <lkocman> also if somebody covers my expenses, I can do some presentations and hands on in Carribean, South Africa, Panama, and Australia
15:17:48 <maxamillion> looks like I missed the whole meeting :/
15:18:40 <langdon> lkocman, thanks for volunteering.. make sure you have cards
15:18:49 <langdon> maxamillion, we have been a bit light today
15:18:54 <nils> maxamillion: you didn't miss awfully much so far, basically that once we fix rida/the FM orchestrator, it should begin building modules on all (staging) arches
15:18:55 <maxamillion> can modules be built outside the fedora infra yet?
15:19:00 <langdon> maxamillion, did you want to discuss some of your examples of parallel install?
15:19:17 <maxamillion> langdon: huh?
15:19:25 <lkocman> maxamillion, copr team claims that they're blocked by api which gives them paths to module repos (base-runtime) so they can add it as repo to chroot
15:19:33 <langdon> maxamillion, scrolling up you might see we can't even do them in fedora-stg at the moment.. asamalik is talking to the copr team about integrating with copr
15:19:36 <lkocman> maxamillion, there is a card for it which we didn't put to this sprint
15:19:48 <lkocman> it should have weight (1) since there is code, it just doesn't work
15:19:55 <lkocman> langdon, do you want to add that card?
15:20:14 <nils> "it just doesn't work" can be any size, so (1) is a safe bet :P
15:20:21 <maxamillion> rgr
15:20:24 <lkocman> nils, cmon it's about importing pungi-trees
15:20:24 <maxamillion> thanks for the info
15:20:27 <lkocman> pungi sends message
15:20:27 <langdon> lkocman, asamalik was going to find out if the copr team could do the work and us just monitor... if that is the case, then it can go in
15:20:29 <lkocman> how hard is that
15:20:30 <asamalik> lkocman: yes please, we should add it
15:20:34 <nils> lkocman: j/k :)
15:20:37 <lkocman> nils, ;-)
15:20:38 <asamalik> langdon: that would work
15:21:03 <lkocman> also then we don't have base runtime do we? :-)))
15:21:11 <asamalik> lkocman: detail :)
15:21:26 <langdon> maxamillion, you said in the workshop at flock that you have lots of examples of wanting the installation of multiple things.. especially different versions of things.. and i wanted to follow up to understand your examples /cc sct
15:21:35 <lkocman> asamalik, langdon well they can build on top of what we build on ... I can make that happen
15:21:38 <maxamillion> lkocman: ohhhhh, right ok
15:21:40 <langdon> lkocman, a hacked one is fine..
15:21:47 <lkocman> like there is a base_runtime tag in koji
15:21:53 <maxamillion> I was mostly curious about being able to use something like mock for building modules on my laptop
15:21:56 <lkocman> maxamillion, base_runtime-1.0-0 in koji.stg
15:22:06 <maxamillion> lkocman: cool
15:22:33 <asamalik> maxamillion: I gues mock with an empty chroot and repo with that base runtime would do
15:22:34 <langdon> maxamillion, i have been trying to keep the team focused on making sure the "cloud" version worked really well first.. then work on local building
15:22:46 <nils> heh "cloud"
15:22:51 <maxamillion> however, rounding back to the concept of parallel installs ... multiple versions of almost every high level programming language is a very common desire
15:23:42 <maxamillion> langdon: I find that odd and backwards, but that's fine ... we don't have to agree on that
15:23:56 <langdon> maxamillion, yeah.. so does most of the team :)
15:24:37 <lkocman> copr wants to read that location of repo from some stable api
15:24:39 <lkocman> such as pdc
15:24:41 <maxamillion> well at least I'm crazy in good company? ... or something
15:24:43 <nils> we gotta do one thing and the other, I don't care much about order
15:24:43 <lkocman> and we already do offer that call
15:24:49 <langdon> maxamillion, my argument is because if people have to use the "cloud version" it will get much better much faster.. and that will be the default entry point for people.. if they can fall back to a different set of tools (ie. local) then it will slow the pass of improvement
15:24:53 <lkocman> just don't fill pdc due listening to incorrect message
15:24:54 <lkocman> that's all
15:25:29 <nils> so, anything we can #info from that?
15:25:44 <maxamillion> langdon: yeah, I could see that happening ... also, please call it "hosted" or something, nothing about that build system is "cloud"
15:25:57 <nils> maxamillion++
15:26:00 <zodbot> nils: Karma for maxamillion changed to 10 (for the f24 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
15:26:09 <langdon> maxamillion, ha.. first word that popped in to my head.. i am gonna say "cloud" all the time to make you angry ;)
15:26:20 <maxamillion> *fine* then ;)
15:26:23 <nils> well, copr is "cloud", at least the URL says so
15:26:57 <nils> so, no info?
15:26:59 <langdon> maxamillion, so.. on parallel.. do you really think most data centers run (want to run) multiple apps on the same server that would require multiple language versions?
15:27:32 <nils> I think that's more of a devel-side requirement
15:27:37 <langdon> nils, i don't really have an info idea on the hosted vs local discussion.. maybe on the copr stuff?
15:28:00 <nils> langdon: no idea, that's why I ask you guys :)
15:28:28 <maxamillion> langdon: today? yes ... in 10 years? I hope not
15:28:34 <langdon> re: parallel) i guess my experience is that people dedicate an instance to the app.. but maybe that is a symptom of the "no parallel install" problem rather than a typical use case
15:28:41 <asamalik> nils: copr team needs the api to get the base runtime repo, and then they can work on building modules
15:28:43 <langdon> maxamillion, yeah.. thats what i wonder too
15:29:08 <asamalik> nils: they will do the work themselves, someone (I?) will observe
15:29:38 <nils> langdon: asamalik's comments ^^, sound good for #info?
15:29:41 <maxamillion> langdon: majority of large companies I've interacted with aren't even running on RHEL7 yet because it's "too new" much less even looking at anything like docker ... they are "evaluating" RHEL7 or possibly just starting to use it in a limited capacity
15:30:08 <langdon> nils, first one yes.. 2nd one, meh
15:30:09 <maxamillion> langdon: just to level set the kind of mind set you're going to be dealing with ... so hosting a bunch of things on a single system is not out of the norm
15:30:32 <nils> langdon: okie
15:31:02 <langdon> lkocman, asamalik do you think you guys could info what needs to be done to enable the copr backend for rida?
15:31:22 <nils> #info the COPR team needs the API to get the base runtime repo, then they can start working on building modules
15:31:45 <langdon> maxamillion, so.. i also think it varies by business size.. (as well as like startups vs established).. but.. let me table this til we finish sorting the copr conversation
15:32:06 <langdon> #topic alternate backends for rida (module build orchestrator)
15:32:31 <nils> langdon: do we #topic in open floor?
15:32:33 <langdon> ok.. lkocman and asamalik have a list of things that need to be done to enable using copr instead of koji as the backend for rida
15:32:46 <langdon> nils, i just switched to a dedicated topic :)
15:32:52 <nils> heh
15:32:53 <langdon> now they will info them ;)
15:33:10 <nils> langdon: should the COPR #info above go into the topic?
15:33:12 <langdon> then i think we could switch to parallel for a bit.. then back to open floor
15:33:25 <langdon> nils, sure? i don't think it hurts either way
15:33:27 <maxamillion> langdon: you brought it up :)
15:33:29 <nils> cool
15:33:34 <nils> gimme a sec to sort things
15:33:39 <nils> #undo
15:33:39 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x1f343bd0>
15:33:39 <langdon> maxamillion, yeah yeah
15:33:42 <nils> #undo
15:33:42 <zodbot> Removing item from minutes: INFO by nils at 15:31:22 : the COPR team needs the API to get the base runtime repo, then they can start working on building modules
15:34:03 <nils> #topic alternate backends for rida (module build orchestrator)
15:34:09 <nils> #info the COPR team needs the API to get the base runtime repo, then they can start working on building modules
15:34:17 <langdon> i was hoping they ( asamalik & lkocman ) would info card links
15:34:46 <langdon> as well
15:35:28 <asamalik> langdon: there is a card about research about copr building modules
15:35:38 <asamalik> which I almost finished last sprint
15:35:52 <asamalik> but it's in new, because it wasn't a priority :)
15:36:16 <asamalik> let me find it
15:36:32 <langdon> asamalik,  i guess I am hoping that if we can point the copr team at the nec. cards they might just be able to run with it
15:37:17 <asamalik> https://taiga.fedorainfracloud.org/project/modularity/us/554?kanban-status=470
15:37:45 <asamalik> I should plan it for this sprint to finish it
15:38:05 <langdon> #chair asamalik lkocman
15:38:05 <zodbot> Current chairs: asamalik dgilmore langdon lkocman nils sct tflink
15:38:18 <asamalik> as a backend for rida - copr should technically support everything
15:38:55 <lkocman> #info This particular card is currently blocking copr (import of base runtime repos + location) https://taiga.fedorainfracloud.org/project/modularity/us/839?kanban-status=480
15:39:31 <asamalik> to make building modules for community members easier - for example without the modulemd.yaml (that would be automatically generated by copr) - there needs to be the API
15:39:47 <asamalik> so they can choose "module" as another chroot
15:40:01 <asamalik> and build the same thing for example for f23, f24, and as a module
15:40:04 * lkocman needs to leave folks. Ciao
15:40:08 <asamalik> the same way they are used to
15:40:11 <langdon> asamalik, so.. in a couple hours do you think you could write all the cards that are needed?
15:40:20 <asamalik> langdon: yap
15:40:55 <langdon> asamalik, ok.. why don't we add doing that to the sprint and then talk to the copr folks about implementing them
15:41:12 <langdon> asamalik, including the one lkocman info'd above
15:41:19 <nils> lkocman: see ya
15:41:49 <asamalik> langdon: agree
15:41:57 <asamalik> langdon: I'll do the cards
15:42:05 <langdon> asamalik, cool
15:42:22 <nils> asamalik: you can "#action asamalik write cards for ..."
15:42:25 <nils> :)
15:42:43 <nils> otherwise I'll do it for you, but it'll be plenty unspecific :P
15:42:52 * dgilmore notes anything built in copr can not be called fedora
15:43:06 <langdon> dgilmore, +1 but it makes adoption easier
15:43:10 <asamalik> #action asamalik write cards for builds in copr
15:43:20 <nils> asamalik++
15:43:36 <asamalik> dgilmore: +1, that would be "community" modules - the same way as copr repos are now
15:43:43 <nils> #info anything built in copr can not be called fedora
15:44:05 <langdon> ok.. should we talk about parallel install more? or just back to open floor?
15:44:18 <asamalik> dgilmore: also, scratch builds can be done in copr, as it supports multiple packages with the same NVR + dist tag (one per project=module)
15:44:50 <nils> langdon: I have one thing for open floor that should only take a minute, namely that the creative types probably want a decision on the badge
15:45:18 <langdon> ok.. i think we can table the parallel install convo anyway.. we need sct to write his doc!! :)
15:45:24 <dgilmore> langdon: thats fine, you need to make new trademarks for whats shipped
15:45:25 <langdon> #topic open floor (again)
15:45:51 <nils> #info Badges: https://fedorahosted.org/fedora-badges/ticket/466
15:45:53 <dgilmore> asamalik: scatch builds in koji can be done over and over as well
15:45:58 <sct> langdon: I hear you. :)
15:47:07 <langdon> sct, :)
15:47:17 <langdon> ok.. nils.. whats the next step on badges?
15:47:21 <nils> guys, do you think that the badge ticket above can be voted on?
15:47:42 <langdon> ohh.. i hadn't seen latest
15:47:54 <nils> langdon: there are two suggested badges, one "why not average them" from me and not much commo since
15:48:09 <nils> I mean, two suggestions that only slightly differ in color
15:48:32 <nils> To me personally, the one with dark green looks... off and the one with white border is to hard on the retina.
15:48:43 <nils> *too hard
15:49:01 <langdon> umm.. i have almost no opinion.. they all look way better than i can do :)
15:49:11 <nils> haha, that's why we don't do it  :)
15:49:21 <asamalik> dgilmore: I heard that koji can't handle multiple packages with the same NVR + dist tag, so scratch builds of modules might be a problem... and I am not a Koji expert myself... so that's not true? that would be great, if we could keep copr just for the community stuff
15:49:47 <nils> asamalik: well, it would require that koji built all the packages scratch, too
15:49:59 <nils> AIUI it doesn't really care about scratch packages' NVRs
15:50:10 <langdon> sure.. want to vote with whoever is here? I am not sure the results require a significant number of people to be present.. its just the look of a badge right? or, send it to the ML for a vote?
15:50:15 <sct> asamalik: I think having multiple builds with the same ID is just a fundamentally bad idea anyway from a tracking perspective
15:50:20 <nils> langdon: no quorum I'm afraid
15:50:35 <nils> langdon: but I wanted to raise it, so maybe you guys can have a quorum next time ;)
15:50:51 <langdon> nils, docs say lazy consensus .. so .. we can send to the ml and call it done
15:51:02 <asamalik> nils: ha! so that would be fine
15:51:02 <nils> langdon: will you do?
15:51:06 <asamalik> langdon: I can vote!
15:51:14 <langdon> nils, yeah.. i will
15:51:24 <nils> asamalik: you don't have voting privileges, sorry (neither do I)
15:51:35 <asamalik> sct: I mean just for the scratch builds... generally, I agree
15:51:42 <langdon> nils, we should probably review thta based on the outcome of the badge vote
15:51:44 <asamalik> nils: :(
15:51:44 <nils> #action langdon pings voting WG members on ML about badge vote
15:51:54 <langdon> ok.. another FYI..
15:52:03 <sct> asamalik: Even then, I think there's a lot of potential for confusion
15:52:18 <sct> you still want to be able to test scratch builds etc
15:52:21 <langdon> the server wg: <sgallagh> #agreed Server SIG meeting time moves to Tuesdays at 4pm US/Eastern (following US DST) (+5, 0, -0)
15:52:41 <langdon> i hope some of us will start to attend that meeting regularly
15:52:42 <nils> langdon: so they have the slot before us now? cool
15:52:57 <langdon> nils, no.. Eastern
15:53:04 <nils> ah
15:53:05 <nils> ugh
15:53:13 <langdon> 4h ish from now
15:53:18 <nils> I'm checking my time zone privilege already :o)
15:53:29 <nils> anyway, not anymore conflicting with us
15:53:47 <langdon> #info <sgallagh> #agreed Server SIG meeting time moves to Tuesdays at 4pm US/Eastern (following US DST) (+5, 0, -0)
15:54:10 <langdon> #info hoping modularity team members will be able to start to attend the server-wg meeting now that it doesn't conflict
15:54:40 <langdon> i wonder if this means we could move from the hinterlands of #-3 closer to #fedora-meeting?!?! :)
15:54:47 <geppetto> langdon: Might help if you put it on our calendar
15:54:55 <nils> langdon: hah!
15:55:00 <langdon> geppetto, yeah.. just got word 10m or so ago
15:55:14 <geppetto> langdon: Or someone ... I can do it if you want
15:55:16 * geppetto nods
15:55:34 <nils> #action geppetto puts server SIG meeting time into team calendar
15:55:36 <langdon> geppetto, sure.. i would have already if i didn't have to go dig up which channel it was in :)
15:55:46 <nils> geppetto: ^^ good?`
15:56:33 <geppetto> nils: ha
15:56:54 <nils> I take that as joyful laughter :)
15:57:14 <langdon> :)
15:57:18 <nils> anything else or are we done?
15:57:24 <langdon> i was just typing that
15:57:56 <nils> good
15:58:05 <nils> Thanks for attending, everybody!
15:58:08 <nils> #endmeeting