15:01:22 <bowlofeggs> #startmeeting Bodhi stakeholders (2017-11-06)
15:01:22 <zodbot> Meeting started Tue Nov  7 15:01:22 2017 UTC.  The chair is bowlofeggs. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:22 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:01:22 <zodbot> The meeting name has been set to 'bodhi_stakeholders_(2017-11-06)'
15:01:22 <bowlofeggs> #meetingname bodhi_stakeholders
15:01:22 <zodbot> The meeting name has been set to 'bodhi_stakeholders'
15:01:23 <bowlofeggs> #topic salutations
15:01:23 <bowlofeggs> #chair bowlofeggs dgilmore masta mboddu nirik pbrobinson puiterwijk Kellin jcline
15:01:23 <zodbot> Current chairs: Kellin bowlofeggs dgilmore jcline masta mboddu nirik pbrobinson puiterwijk
15:01:49 * mboddu kinda here
15:06:04 * nirik is here, but on first cup of coffee.
15:06:58 * bowlofeggs is also on first coffee
15:07:10 * puiterwijk is on 0th cup of coffee
15:07:10 <bowlofeggs> i don't have a lot on the agenda today
15:07:23 <bowlofeggs> #topic announcements and information
15:07:23 <bowlofeggs> #info Bodhi 3.0.0 has been released and deployed. It uses Pungi to mash instead of mash.
15:07:23 <bowlofeggs> #info Bodhi 3.1.0 is planned for release "soon", after the freeze. https://github.com/fedora-infra/bodhi/pull/1961
15:07:37 <bowlofeggs> (that PR is the release notes)
15:08:00 <bowlofeggs> 3.1.0 actually has a lot of oldish patches in it
15:08:35 <bowlofeggs> due to the long "winter" of f27 freezes, and due to me wanting modular mashing to have its own release, i basically held lots of patches on the develop branch hostage
15:08:41 <bowlofeggs> some go back to august
15:08:50 <bowlofeggs> so 3.1 will be a largish diff
15:09:01 <bowlofeggs> but it's got some nice changes
15:09:20 <dustymabe> .hello2
15:09:21 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
15:10:11 <dustymabe> bowlofeggs: cool
15:10:13 <bowlofeggs> one of the big things i'm happy about is adding a section to the docs that covers bodhi's server settings
15:10:34 <bowlofeggs> oh and e-mails should once again say the version of the update
15:10:37 <bowlofeggs> that bug annoyed me
15:11:41 <bowlofeggs> i don't have any big items planned to discuss for today other than those announcements, so we can just do open floor
15:11:44 <bowlofeggs> #topic open floor
15:12:48 <dustymabe> I do have a question but I don't know if it is meeting worthy
15:12:54 <bowlofeggs> go 4 it
15:13:15 <dustymabe> I see there is a commit in bodhi for allowing extra pungi command line arguments
15:13:22 <dustymabe> pungi.extracmdline
15:13:23 <bowlofeggs> indeed
15:13:28 <bowlofeggs> patrick added that
15:13:35 <dustymabe> what arguments do we currently pass to bodhi?
15:13:48 * bowlofeggs digs up ansible cgit
15:14:18 <dustymabe> i have it up. what file in ansible cgit are they in?
15:14:29 <bowlofeggs> https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/roles/bodhi2/base/templates/production.ini.j2#n124
15:14:39 <puiterwijk> pungi.extracmdline = --notification-script=/usr/bin/pungi-fedmsg-notification
15:14:58 <dustymabe> cool
15:15:22 <dustymabe> so lsedlar recently added another script that we use to make sure ostrees get signed before moving forward in a compose
15:15:47 <dustymabe> right now the updates runs don't make other artifacts other than ostrees and yum repos
15:16:06 <dustymabe> but if they did I'd really like for us to run that other script that makes sure the ostrees get signed before moving on
15:16:11 <puiterwijk> The pungi-wait-for-ostree-signed-handler you mean?
15:16:14 <dustymabe> puiterwijk: yes
15:16:26 <puiterwijk> Yeah, I had seen that.
15:16:34 <dustymabe> is there any way we could add that?
15:17:00 <puiterwijk> Sure, but let's do that post-freeze, since we're not going to add the AH image generation to there until then probably
15:17:08 <dustymabe> puiterwijk: cool
15:17:14 <bowlofeggs> sounds fine to me too
15:17:19 <dustymabe> yeah I'm fine with waiting, I just didn't want to drop it
15:17:19 <bowlofeggs> agreed on post freeze
15:17:22 <dustymabe> and I thought of it today
15:17:33 <puiterwijk> dustymabe: could you create an infra ticket so we don't forget?
15:17:39 <dustymabe> puiterwijk: yes
15:18:00 <dustymabe> thanks!
15:18:02 <bowlofeggs> any other topics anyone would like to discuss?
15:18:10 <puiterwijk> bowlofeggs: Bodhi 4?
15:18:13 * puiterwijk hides
15:18:36 <bowlofeggs> hahaha
15:18:39 <puiterwijk> Well, at least a headsup that bodhi 4 might be here sooner than Bodhi 2 was to bodhi 3
15:18:44 <bowlofeggs> well, yeah, we probably will do a bodhi 4 at some point too
15:19:04 <dustymabe> major features/changes being?
15:19:13 <bowlofeggs> auth API change is the main thing
15:19:29 <dustymabe> from what to what
15:19:35 <bowlofeggs> i've got a few other backwards incompatible changes i'd like to sneak in too
15:19:39 <puiterwijk> From OpenID to OpenID Connect/OAuth2
15:19:46 <dustymabe> cool
15:20:09 <bowlofeggs> the next big feature goal on my radar is adding containers to bodhi
15:20:19 <bowlofeggs> that should be a lot easier than adding modules was (famous last words)
15:20:38 <bowlofeggs> largely because bodhi is now flexible enough to handle multiple types
15:22:22 * nirik thought bodhi 4 would be the re-write in flask
15:22:24 * nirik runs
15:22:24 <puiterwijk> Might be sooner than you'd expect (collective "you")
15:22:33 <bowlofeggs> nirik: i really would like to do that
15:22:37 <puiterwijk> nirik: me and Randy actually played with that idea... :)
15:22:47 <bowlofeggs> nirik: esecially after the painful patch i wrote to get bodhi working with cornice 2
15:22:59 <bowlofeggs> it took me 3 days to get bodhi to work with cornice 2
15:23:11 <bowlofeggs> and cornice's latest release is 3, which is also backwards incompatible
15:23:48 <bowlofeggs> there's actually a refactor that's even higher priority in my mind - bodhi's state machine code is just awful
15:24:00 <bowlofeggs> and it leads to constant bugs whenever we change it
15:24:13 <bowlofeggs> i really want to implmenent it with a nice state transition function
15:24:24 <puiterwijk> bowlofeggs: so, just a thought you probably already had: let's make the state machine independent of the framework.... That'd make it easier to transition in the future
15:24:26 <bowlofeggs> right now, it's just code that's spread all over the place
15:24:32 <puiterwijk> (if we still want to)
15:24:33 <bowlofeggs> oh of course
15:24:48 <bowlofeggs> the state machine is actually just going to be a post-commit hook for sqlalchemt
15:25:07 <bowlofeggs> so the things that want to change the state actually can't do anything other than signal inputs
15:25:18 <bowlofeggs> the state will transition automatically as appropriate
15:25:32 <bowlofeggs> this way nothing that interacts with updates has to do anything special
15:25:59 <bowlofeggs> i.e., the comment API just adds a comment, and the comment function will not have to check to see if the karma reached a threshold - that'll happen automatically on commit
15:26:43 <bowlofeggs> and so all the places that currently affect the state of the update will be simple, and there will be exactly one place the states transition
15:26:50 <bowlofeggs> which will make this so much easier to maintain
15:27:04 <bowlofeggs> https://github.com/fedora-infra/bodhi/issues/1930 was tempting me to go ahead and do that now
15:27:10 <bowlofeggs> but i have a quicker fix almost ready
15:28:08 <bowlofeggs> anyways, if there are no more thoughts i'm prepared to tell zodbot to END
15:30:12 <bowlofeggs> #endmeeting