14:00:31 <tflink> #startmeeting fedora-qadevel
14:00:31 <zodbot> Meeting started Mon Aug  7 14:00:31 2017 UTC.  The chair is tflink. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:31 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:00:31 <zodbot> The meeting name has been set to 'fedora-qadevel'
14:00:36 <tflink> #topic roll call
14:00:43 <tflink> #chair kparal
14:00:43 <zodbot> Current chairs: kparal tflink
14:02:17 * kparal is here
14:03:37 <tflink> #chair jskladan
14:03:37 <zodbot> Current chairs: jskladan kparal tflink
14:03:46 <tflink> no lukas?
14:04:14 <jskladan> as if Lukas ever used irc...
14:04:29 * tflink shrugs
14:04:35 <tflink> all the shorter of a meeting :)
14:04:42 <tflink> #topic Announcements and Information
14:04:51 <tflink> #info ci-listener is up and running in the CentOS CI Infra - tflink, pingou
14:05:00 <kparal> he has to feed the rabbits, as usual
14:05:00 <tflink> any other items?
14:05:09 <tflink> convenient timing
14:05:17 <kparal> right!
14:05:22 <kparal> I'm starting to suspect something
14:05:27 <kparal> I talked to him and have some info, at least
14:06:12 <jskladan> I'm certain I added something to the wiki earlier this day
14:06:20 <kparal> you added it to agenda it seems
14:06:26 <jskladan> ah, my bad
14:06:27 <kparal> so now we need to discuss it
14:06:30 <tflink> :)
14:06:31 <jskladan> yay
14:06:40 <jskladan> sorry, though
14:07:00 <tflink> #info Moved tickets from Phabricator to the respective repositories https://lists.fedoraproject.org/archives/list/qa-devel@lists.fedoraproject.org/thread/EEA7DJUICDMSCI4R7I54GBEKBMHDBU2X/ - jskladan, kparal
14:07:00 <tflink> #info Archived differentila revisions' "html snapshots" and raw diffs. Accessible at https://fedorapeople.org/groups/qa/phabarchive/differentials/phab.qa.fedoraproject.org/ - jskladan, kparal
14:07:07 * tflink didn't see them at first
14:08:02 <tflink> any comments/questions/additions?
14:08:04 * jskladan has nothing to add
14:08:17 <kparal> nope
14:09:19 <tflink> cool, moving on
14:09:24 <tflink> #topic Phabricator Migration
14:09:44 <tflink> other than the wiki, is everything moved off of phab?
14:10:19 <jskladan> Tickets and Diffs are
14:10:31 <kparal> jskladan: what about the infra diff?
14:10:40 <tflink> other than parts of the wiki, what is left?
14:10:55 <jskladan> kparal: there is no sensible way of moving it, IMO
14:11:07 <kparal> just create a ticket and attach diff
14:11:13 <kparal> or include it as a code block
14:11:19 <jskladan> no me gusta :)
14:11:31 <kparal> jskladan: put the link here so that tflink knows what we're talking about
14:12:20 <jskladan> pff /me searches
14:12:31 <jskladan> #link https://phab.qa.fedoraproject.org/D1208
14:12:53 <jskladan> just so the context is complete - there is no _good_ way of doing stuff like this now
14:12:54 <kparal> the point is, a PR can't be created from a file
14:13:00 <tflink> yeah, I don't know of a good way to do that other than sending patches around
14:13:14 <jskladan> aka "PR for a repo, that is not on pagure, but we'd still like to discuss it"
14:13:29 <tflink> review on ML
14:13:34 <kparal> so just upload somewhere where you can show differences nicely, and link it in an issue
14:13:40 <kparal> tflink: noooo
14:13:51 <jskladan> kparal: could we do it with all the pr's?
14:13:54 <jskladan> :)
14:14:06 <tflink> jskladan: it would be consistent ....
14:14:14 <jskladan> tflink: +1
14:14:25 <tflink> but I, for one, am lazy enough that a nice visual representation is appreciated
14:14:51 <kparal> maybe pagure can highlight patches as code blocks?
14:14:57 <kparal> we can try
14:15:27 <jskladan> if it can, I'll move the pagure's score from -1345.3 to -1345.2! :)
14:16:37 <kparal> nope, it can't
14:17:16 <tflink> #info for now, any patches on outside projects (i.e. infra) will have to be discussed as patches, either in tickets or on ML
14:17:40 * kparal objects to ML
14:18:08 <kparal> also, we don't have a good way to run linter for our projects right now
14:18:14 <kparal> I'll file a ticket about it
14:18:42 <tflink> pagure CI or maybe travis ci is likely the most sane way forward
14:19:07 <kparal> I don't want to wait on CI to find out I have a linter error
14:19:13 <kparal> of course it's good to have something server-side
14:19:21 <kparal> but I want to also have something client side
14:19:23 <tflink> doit?
14:19:34 <kparal> sure, somebody needs to implement it
14:19:53 <kparal> on the diff only, not on the whole project. (I can run "flake8 .")
14:20:02 * tflink can put it on his todo list
14:20:20 <tflink> yeah, that might take some work
14:20:34 <jskladan> can git do pre-commit check?
14:20:48 <kparal> yes, that's one way. I already looked at it a bit
14:21:10 <kparal> it just needs some testing and figuring out corner cases
14:21:18 <kparal> like if you can override a pre-commit hook for example
14:21:21 <kparal> I don't know yet
14:21:39 <kparal> anyway, that was just a side-note
14:21:54 <jskladan> pre-commit hook would be nice, imo
14:22:09 <jskladan> although, I can see it becoming quite a pita fast
14:22:28 <jskladan> (or I'll just learn to write pep8 compatible code on the first try :))
14:23:23 <kparal> right, for WIP branches that might be annoying
14:23:25 <jskladan> how about autopep8 as post-push hook on server?
14:24:03 <kparal> no likey
14:24:08 <jskladan> kparal: why not?
14:24:23 <kparal> I don't want my code to be reformatted randomly
14:24:28 <jskladan> I for one do not see a reason against using autopep8
14:24:38 <jskladan> then it's fine
14:24:40 <kparal> let's skip this discussion right now
14:24:41 <jskladan> it is not random :)
14:25:15 <tflink> #info now that we don't have phab linting our code, we need to find a new way to have that check run
14:26:11 <kparal> #link https://pagure.io/taskotron/issue/227
14:26:26 <tflink> thanks
14:26:40 <tflink> what do we want to do about the wiki pages? fp.o/wiki ?
14:26:58 <jskladan> #info now that we don't have phab running unittests on our code, we need to find a new way to have that check run
14:26:59 <kparal> throw them away?
14:27:12 <kparal> jskladan: unit test are working with "py.test" command
14:27:40 <kparal> tflink: the only question I have is regarding this: https://phab.qa.fedoraproject.org/w/blockerbugs/building-for-infra/
14:27:50 <kparal> all of the other pages are outdated anyway, I think
14:28:07 <tflink> unless we want to save them for any archival purposes
14:28:16 <jskladan> kparal: linter is working with "flake8" command
14:28:20 <kparal> I don't feel much need
14:28:28 <tflink> that particular page looks like it should move to the fp.o wiki
14:28:48 <tflink> #action tflink to migrate blockerbugs "building for infra" page from phab wiki to fp.o wiki
14:28:56 <kparal> tflink: or blockerbugs git. do we have some pages with similar SOPs?
14:29:08 <tflink> hrm, that's a good point
14:29:11 <tflink> #undo
14:29:11 <zodbot> Removing item from minutes: ACTION by tflink at 14:28:48 : tflink to migrate blockerbugs "building for infra" page from phab wiki to fp.o wiki
14:29:37 <kparal> jskladan: the test suite is run always as a whole. but for linters you only want to see your changes. running "flake8 ." is not helpful (try it)
14:30:19 <tflink> it'd probably be better to put in a README or in the docs that technically already exist for blockerbugs
14:30:26 <tflink> no idea how badly out-of-date they are, though
14:30:27 <jskladan> kparal: well, then our code should be pep8 valid to start with, no? :) (not that I don't get what you try to say, but still)
14:30:56 <kparal> jskladan: no it's not, because that would need somebody to make it pep8 valid first
14:31:03 <tflink> unless we start explicitly excluding the violations that don't matter to us
14:31:07 <kparal> jskladan: you're welcome to do that, though
14:31:19 <jskladan> tflink: we already exclude quite a lot of violations
14:31:24 <jskladan> kparal: ok, will be done tomorrow
14:31:27 <tflink> I mean line-by-line
14:31:44 <kparal> jskladan: also include flake8 output then
14:31:54 <jskladan> kparal: ?
14:31:58 <kparal> and I don't think there's autoflake8 :)
14:32:21 <tflink> either fixing the existing violations or excluding each occurrence sounds like a way forward
14:32:36 * tflink is a little concerned about how much time that could take
14:32:48 <tflink> but I haven't tried to lint all of our codebases for a while
14:33:47 <kparal> I think we're misunderstanding each other. there's no point in trying to fix existing linter warnings. a lot of effort for little gain. the reasonable way forward is to enforce valid linting on new patches, as we've done for the past year or two
14:34:01 <kparal> but for that, we need a tool. phab used to do that for us
14:34:09 <tflink> I don't think that we're misunderstanding eachother
14:34:22 <tflink> I'm not sure that the tool you talk about is that much less work than just fixing the errors
14:35:04 <kparal> are you going to fix even all files in the test suite?
14:35:36 <kparal> and are so many lines changed in the git history worth it?
14:35:37 <jskladan> kparal: testsuite can IMO easily be excluded from lint check
14:35:52 <jskladan> kparal: we had huge diffs before..
14:36:18 <kparal> it can be excluded, but then you're back at something running the check, like "make lint"
14:36:52 <jskladan> how is that different from running 'flake8'?
14:37:02 <jskladan> kparal: I don't really see the problem, honestly
14:37:03 <tflink> which can be 'flake8 .' instead of some way of figuring out the diff and running the linter against that diff
14:37:24 <tflink> without phab, I can't think of a good way of reliably determining what's changed
14:37:44 <kparal> make lint develop, diff against develop
14:37:55 <kparal> or make lint staged, run against git diff --staged
14:38:03 <kparal> I've seen some oneliners to do that
14:38:15 <tflink> sounds like we have a volunteer to figure out the tooling there
14:38:20 <jskladan> :)
14:38:23 <kparal> I already assigned the ticket to myself
14:38:26 <tflink> ok
14:38:34 <tflink> if you can find a decent way to do it, I'm not against the idea
14:38:36 <kparal> and I'm trying to tell you that we've spent too much time on this already
14:38:42 <tflink> fair enough
14:39:05 <tflink> any other thoughts/comments/concerns about the migration?
14:39:29 <kparal> I've tried to replace all the links and notes about phab
14:39:35 <kparal> if I've missed something, please fix
14:39:41 <jskladan> kparal: the amount of slack in your vigorous rigor, sometimes surprises me :)
14:39:50 <kparal> I don't know about anything else that's missing from the migration
14:39:57 <jskladan> neither do I
14:40:06 <jskladan> we have not moved any of the files/attachments
14:40:17 <jskladan> but I could not find a way to easily find those in phab
14:40:19 <tflink> are they needed?
14:40:28 <tflink> IIRC, they're stored in the FS
14:40:40 <jskladan> I don't think so, just that we discussed it
14:40:56 <jskladan> and the "philes" (or whatever they call it) thing is full of nonsense
14:41:42 <tflink> yeah, I'm not seeing a quick way to get them from the fs
14:41:49 <tflink> do we need them?
14:43:26 <jskladan> nope
14:43:33 * tflink takes silence and "nope" as nope
14:43:36 <tflink> :)
14:43:49 <jskladan> btw: `git diff HEAD^ | flake8 -diff` (or `git diff HEAD | flake8 -diff` for unstaged changes)
14:44:06 <kparal> jskladan: yes, but you need to exclude non-python files
14:44:11 * tflink wants to go through the wiki to make sure that everything useful has been moved
14:44:25 <tflink> also, how do you do stuff that's already been committed?
14:44:40 <jskladan> tflink: ?
14:45:07 <tflink> if I have a feature branch that has multiple commits - I don't think that command would work
14:45:22 <tflink> unless I'm forgetting how HEAD^ works
14:45:46 <kparal> yes, but you can replace it with develop
14:46:02 <jskladan> you could do develop or HEAD~10
14:46:27 <tflink> which is more appropriate, develop or origin/develop?
14:46:38 <kparal> that depends on what you want :)
14:46:42 <tflink> nvm, I'm sending this conversation into a rabbit hole
14:46:49 <kparal> it is
14:46:57 <jskladan> kparal: also, maybe you should try it first, since your computer really usually works differently than min, but the command really only checks python files
14:47:01 <tflink> anything else on this?
14:47:02 <jskladan> so..
14:47:03 <jskladan> whatever
14:47:17 <kparal> jskladan: it seems mine does :)
14:47:38 <kparal> tflink: can't think of anything else
14:48:04 <tflink> ok, moving on
14:48:07 <tflink> #topic Tasking
14:48:15 <tflink> what are folks planning to work on this week?
14:48:42 <jskladan> tflink: argue with kparal about pep8, it seems :D
14:48:49 <jskladan> other than that, my platter is empty
14:48:51 * tflink is planning to help with getting stg running again and the ansiblize branch of libtaskotron
14:49:23 <kparal> has anyone looked at https://github.com/fedora-infra/bodhi/pull/1694 yet?
14:49:27 <kparal> just wondering
14:49:30 <tflink> oh, I forgotabout upstreamfirst
14:49:33 <tflink> sigh
14:49:49 <kparal> I was planning to look at it
14:49:57 <kparal> but migration took priority
14:50:44 * tflink has not looked at it
14:50:48 <jskladan> me neither
14:51:05 <kparal> so I'm planning to look at this one, at 'no more alpha' proposal from adam, and help to figure out what's wrong with taskotron-stg if needed
14:51:08 <tflink> upstreamfirst needs to be upgraded to newest pagure and the status app that roshi hasn't been deployed yet
14:51:21 <kparal> also I guess I'll need to package the ansiblize version of libtaskotron?
14:51:26 <tflink> trigger is fubared, AFAIK
14:51:31 <tflink> on stg
14:51:38 <tflink> it's spewing lots of TBs to journal
14:52:11 <kparal> tflink: what's the difference between https://upstreamfirst.fedorainfracloud.org/ and http://pkgs.fedoraproject.org/ with tests/ subdirectory?
14:52:52 <tflink> kparal: upstreamfirst is a staging area, isn't meant to be long term
14:53:00 <jskladan> tflink: that really is weird... so it sounds like reinstalling the triger/fedmsg-hub is the next step?
14:53:12 <tflink> from the TBs, that owuld make sense
14:54:05 <kparal> tflink: also, I learned why nfs works on dev
14:54:19 <kparal> martin simply turned selinux off, turned the service on, and reenabled selinux
14:54:21 <tflink> oh?
14:54:25 <kparal> it will break after reboot
14:54:25 <tflink> ha
14:54:27 <kparal> problem solved :)
14:54:52 * tflink thought martin knew better than to just do stuff like that :(
14:55:01 <kparal> it might be a good idea to have a file with workarounds to current problems :)
14:55:04 <tflink> sounds like we have work to do on dev/stg, then
14:55:18 <kparal> at least that's what lbrabec told me, I wasn't speaking to martin directly
14:55:40 <tflink> so to rephrase my plans for the week: taskotron-stg, upstreamfirst (pagure, uf-monitor), then libtaskotron's ansiblize branch
14:56:03 <tflink> losing folks makes such a positive impact on velocity, doesn't it?
14:56:29 <kparal> nobody is slowing us down!
14:56:36 <kparal> that's what you meant, right?
14:56:44 <tflink> I suppose that's one way to look at it
14:57:02 <tflink> anyhow, I'm just whining at this point
14:57:03 <jskladan> I was also quite surprised, that doing manual overrides, instead of putting it to ansible, was supposedly "the way"
14:57:07 <jskladan> but who am I to argue
14:57:09 <tflink> it's not
14:57:33 <tflink> the only time that's OK is if prod is broken and we need to get it working while a better fix is found
14:57:34 <jskladan> good to know, then
14:58:05 <tflink> or if we're waiting on a fix to make it into the repos
14:59:14 <tflink> anything else?
14:59:23 * tflink is surprised that this took the whole hour
14:59:36 <kparal> a few general questions regarding ansiblize, perhaps
14:59:54 <tflink> kparal: good for a meeting or ML, IRC later?
15:00:06 <kparal> either way
15:00:21 <kparal> let's close this and continue on fedora-qa
15:00:23 <tflink> IRC later, then.
15:00:28 <tflink> sounds good to me
15:00:32 <tflink> #topic Open Floor
15:00:37 <kparal> nothing here
15:00:39 <tflink> Any other topics that you all want to bring up?
15:00:46 <tflink> for the three of us, anyways :-/
15:01:28 * tflink lights the fuse
15:01:39 <tflink> thanks for coming, everyone
15:01:43 * tflink will send out minutes shortly
15:01:45 <tflink> #endmeeting