mozilla :: #conduit

11 Aug 2017
00:49mcotebah have to delete and re-add
00:50mcoteahaha and since I renamed the old rtd project, thinking it would change the url slug, even after I deleted it it tells me the name is in use
00:50* mcote hopes the system coalesces at some point
03:14dkl|afkdavidwalsh: i have had to restart the whole thing when making hanges to the extension code. 'docker-compose stop; docker-compose up'
03:15dkl|afki have not had to do arc liberate when adding extension code but seems to be needed when it is deployed to production
03:16davidwalshdkl|afk: Literally just it out; `use Bugzilla::Constants;` was missing, so `my $user = Bugzilla->login(LOGIN_REQUIRED);` was bombing
03:20dkl|afkah yep
03:46globimadueme: it would be simpler to grab it from s3 directly, as well as ensuring what is landed is what is requested to be landed. having transplant request from lando adds another dependency and another point of failure.
03:49davidwalshglob: Do I need my own `throw` statement if I use user->check?
03:49globdavidwalsh: no
03:50davidwalshglob: Do you know what status code that throws?
03:51globdavidwalsh: object_does_not_exist iirc. pretty sure you can override the error thrown. look at the `check` sub in Bugzilla::Object
03:53glob(i don't think i've ever seen that error overridden to be honest)
03:54globyou probably want to use the object cache too
03:54globmy $user = Bugzilla::User->check({ id => $params->{user_id}, cache => 1 })
12:42imaduemeHey guys, I'll be out today. dkl feel free to merge your and davidwalsh's patches on lando-ui whenever, all 3 look great.
12:42imaduemeglob: that makes sense, we'll continue with s3 then :)
13:01dklimadueme: thanks
13:33zalunglob: I've got a wip solution for s3
13:33zalunworking and saving in my private bucket, with tests
14:23mcotegr, readthedocs says that "Mozilla Conduit" still exists
14:23mcoteeven though it 404s when you load it
14:32mcotebah seems they have issues with fully deleting content
14:32mcotemoz-conduit it is, sigh
14:47dklmcote: talk to their support?
14:49mcoteyeah I tried to find some contact info
14:49mcoteseems they just refer everyone to Stack Overflow heh
14:49mcotebut I found some open issues on their GitHub that discuss how they don't fully delete everything
14:49zalunIRC standup today?
14:51zalunstandup: I've been working on lando api - patch in S3 bucket. Made it working, with tests. Refactoring and PR on Monday
14:51mcotestandup: docs :) not complete yet but feel free to have a look
14:53mcoteI think davidwalsh is the only other person around today
14:54dklirc standup: working on build errors with latest phabricator-extensions build. getting close
14:55davidwalshstandup: I have ( in for the new BMO endpoint , I have ( in to use said endpoint; now fixing the BMO PR for whatever nit it's failing on, then back to the phab bug validation
14:58mcotedavidwalsh: wow I can't believe you are already done the endpoint
14:58mcotegood stuff!
14:58mcotedkl: heh I finally got around to installing vidyo on my XPS... is it super tiny for you too? :)
14:58mcoteor is it only because I have an external monitor connected, for some reason?
14:59dklmcote: you have to adjust the hidpi settings for the app
14:59dkleasiest way is to pin vidyo to taskbar...
14:59mcotehm it is pinned...
14:59dklright click on the pinned icon, the right click again on the Vidyo item in the menu and properties
15:00dklthen click on the Compatibility tab
15:00dklclick on override high dpi scaling
15:00dkland make sure it says system in the drop down. Apply
15:00mcoteahh I see
15:01dklnext time you run vidyo it should look normal albeit little fuzzy
15:01davidwalshmcote: I'd have the validation part done too if our environments didn't hate me
15:01dklsome say it helps video performance as well but i am not sure about that
15:02dklmcote: anyway you should be able to do similar to other apps that look bad under hidpi
15:03davidwalshzalun: When someone uses "arc diff", what information comes from arc to identify the user pushing it?
15:05mcotedkl: hm I did all that, but upon launching it, it's still tiny
15:10mcotedkl: I have "Override high DPI scaling behavior" checked, with "Scaling performed by 'System'" set
15:10mcotethat's all I need?
15:15dklshould be. you will need to fully quit vidyo
15:15dklright click the icon in your systray and click exit
15:20mcotenow it looks like the right size on my external monitor but not on my main monitor
15:41zalundavidwalsh: I'm on the phone, I don't remember tbh
15:47gpsthe led to
15:48gpsso that fixes one of the major scaling problems you were experiencing this week
15:50smacleodgps: excellent :) - thanks for the links
15:51smacleodckolos / mcote ^
15:51ckolosoh that's just ducky.
15:53ckolosso we'll need to use a newer version of mercurial in the container as well as bump to this version of the phab code asap
15:57gpsyes, *all* {Mercurial, Git, Subversion} installs at Mozilla should be updated ASAP
15:58gpsanything operating on unsanitized user input is the most important
16:08ckolosdkl: PR coming your way
16:09ckolosgps: is a new version of mercurial coming?
16:11davidwalshHow can it be possible that I get into bmo-extensions phabricator container, change a source PHP file, but refreshing the page where it's used shows no change?
16:12davidwalshzalun: dkl: ckolos: Any idea? It's becoming a blocker that I don't have a phab env where no changes take effect
16:13ckolosdavidwalsh: does the console show that it's being delivered direct or via cloudfront?
16:13circleci-botSuccess: ckolos's build (#63; push) in mozilla-services/mozphab (ckolos/CVE-2017-1000117) --
16:14davidwalshckolos ^
16:14davidwalshAdditional oddity: the "Bugzilla" extension I merged into phabricator-extensions yesterday isn't appearing there either
16:14ckolosoh, sorry, this is local
16:14davidwalshDespite a complete rebuild
16:15ckolosphabricator extensions isn't building new images atm, unless you are also using a local image for that
16:15davidwalshI'd prefer to mount the extensions directory but pages *time out* when I try that
16:15davidwalshTakes *forever*
16:15ckolosif you are using local, turn off caching and rebuild
16:18davidwalshckolos: Turn off caching from phab or the container? And how?
16:19gpsckolos: it was released yesterday
16:19ckoloswhen docker builds the container, you can either leave caching on or turn it off --no-cache
16:20ckolosgps: is still showing 4.2.2 as latest
16:20ckolosoh wait, sorry, that's for windows
16:30circleci-botSuccess: ckolos's build (#64; push) in mozilla-services/mozphab (ckolos/CVE-2017-1000117) --
16:31davidwalshckolos: Even a no-cache build doesn't pull in the newly merged phabricator-extensions dir
16:33ckolosdavidwalsh: sorry, maybe I'm confused about what you're doing.
16:33ckolosYou're building a local stack
16:33davidwalshckolos: Right, from this:
16:33davidwalshSuper vanilla docker-compose
16:33ckolosand in the process of the build, a merged dir isn't showing up in it?
16:33davidwalshUnless `mozilla/phabext` isn't updating
16:34ckolosit is not
16:34ckolosit got broken last night
16:34ckolosb/c the new version of mozphab doesn't have the correct library map
16:34ckolosdkl was working on that afaik
16:35davidwalshSo to skirt that issue, I try to mount my local phabricator-extensions to the phabricator container's extensions dir
16:35davidwalshThe problem there is it takes sooooo long to load each page that it times out
16:35davidwalshBut AFAIK that's a Mac issue
16:36ckolosso as dumb as this sounds
16:37ckolosthe original osx version of docker used a VM as a docker host
16:37davidwalshEven still, if I use vi to edit a fie in the phabricator shows no effect; I'm completely confused
16:37davidwalshAn extension file, that is
16:47dklckolos: it still does it just uses a vm that comes built into osx
16:47dklit used to use virtualbox (Docker Toolbox)
16:47ckolosyeah, that
16:48dklthe new Docker for Mac uses a new filesystem called osxfs for mounting host volumes into containers that, last time i used it, was very slow.
16:48dklpeople have had better success with using nfs instead
16:48dklbut requires some hackery
17:17davidwalshWhew, I found a semi-acceptable solution so I'm no longer blocked
17:20davidwalshckolos: dkl: FYI, people recommending this:
17:20* ckolos uses linux ;)
17:21davidwalshYikes that's a lot of setup
17:21davidwalshckolos: hippie
17:22ckolosI am amused by that statement
17:23davidwalsh"docker-compose down && docker-compose up -d --no-deps --build phabricator && docker-compose up" is my new drug, btw
17:23davidwalshI still can't edit with vi instead container but I'll take this
17:30dklyeah but the only downside is you have a full copy of your data on both sides.
17:30dkldepends on how much data you are keeping in sync
17:33dklhm mruby
17:39* ckolos bbiab
17:58davidwalshdkl: We need to catch up before the day so I can figure out next week
18:24dkldavidwalsh: you bet. let me finish committing the PR that dylan just approved
18:24dylandkl: make sure to edit the commit message on GH
18:24dylanor I can do
18:25dkldylan: need to commit the index nit you mentioned and then I can squash into single commit right?
18:25dylanI'd need to re-approve then, but yeah
18:27dylanmight need to rebase against master first to force the circleci tests to run
18:41dylandkl: want me to do that?
18:49dkldylan: up to you. having an issue with my dev env atm
18:52davidwalshdkl: Can you sanity check me here?
18:55dkldavidwalsh: right. you need my $phab_user = Bugzilla::User->check(...); return { result => $phab_user->can_see_bug(...) };
18:57dylanI recommend not naming it $phab_user as that implies the opposite of what I think you're doing.
19:04dkl$other_user maybe
19:07davidwalshchanged to $target_user
19:15dkldylan: ugh. what is the best way to merge master into my branch without generating the extra merge commit?
19:18dylanrebase? is this about that PR?
19:18mcoteindeed, dkl what are you reviewing?
19:19dkldylan: yes
19:19mcotewhat is it?
19:19dklmcote: not reviewing, trying to update my PR to latest master, make a few changes, submit for last approval and then merge
19:20mcoteregardless, you want to rebase don't you?
19:21mcoteoh this?
19:22mcotecould dylan maybe merge this?
19:22dylanone sec
19:22mcotedkl: most important thing today is to chat with davidwalsh & update him on what you're doing/your plans
19:24dylanthere's some conflicts between #166 and master
19:26dylanoh cool
19:26dylansince this PR is against a branch in the bmo repo, I can push up my fixed version
19:26dylangit diff http-auth-fix origin/http-auth-fix
19:27dylan# nothing is different, but the history is now rebased
19:28dylantests are running now:
19:35dkldylan: quick approvel on
19:37dylanapproved already, but CI has to finish. I'll commit when it does
20:02dylandkl: bug 1382225 can be resolved -- code landed
20:02firebot ASSIGNED, Missing code from PhabBugz extension such as http basic auth support and other minor improvements
20:08dkldylan: thanks
20:31ckolosoh yay, a db upgrade
20:38davidwalshmcote: Spoke with dkl, we're on teh same page
20:39ckolosokay guys, latest version of mercurial/mozphab/phabext running on dev
20:40ckoloswait, wait...
20:41dkluh oh
20:41ckolosno worries ;)
21:18dklso how do you interdiff in github PRs?
21:24dylanI recommend having hub, and then you can add the other person's remote:
21:24dylanhub remote add darkwing darkwing/bmo
21:24dylangit fetch darkwing; git diff darkwing/branch origin/branch
21:25dklhub, that is a command line utility?
21:25dylanI just can't remember what to write before the remote url and hub does that for me
21:25dylanbut basically you can have a bunch of remotes and diff between them
21:26dylanlike a real-ish example: git diff mary/bug-1348439 seban/bug-1381752
21:26firebot FIXED, umohm12 History item "reporter accessible: true" is confusing
21:26firebot FIXED, sebastinssanty Change the layout of the new bug form to be more usable
21:27dklcool will play with it some. just seems odd that u cant do it in the GH UI
21:27dylanmy git remote -v looks like:
21:27dylanI think you can somehow, but the UI is fiddly and I don't know how :)
21:28dylanfound it:
21:29dylanclick "compare across forks"
21:30dklwhen looking at the last commit of the PR, there is a drop down at the top left "Changes from all commits"
21:30dklif you drop it down, there is a choice "Show changes since your last review" but it is disabled.
21:31dklSeems that would do what I want but do not know when conditions have to happen to turn that on.
21:31dkldylan: strange selenium failure on
21:31dkldinner time. see ya
21:33dylandkl|afk: that's the race condition I fixed in test-passwords
21:34dkl|afkah how do you restart it without doing another commit?
21:34dylan, "rebuild"
21:34dylanif you don't see the button I'll prod the org config
21:40dylandkl|afk: if you approve (and rebase against it) the error won't happen.
21:42davidwalshdkl|afk: ckolos: How can we prod this to get updated?
21:43ckolosYou said things in the form of a sentence, but the parser blew up half way through.
21:44* ckolos thinks on the usage of prod
21:44ckolosprod == poke?
21:44ckolos davidwalsh: ^^ ?
21:44ckolosi.e. can I get a new version of that image?
21:45dylanit's an automated build, but it has been failing:
21:45ckolosI don't know anything about that image, so I think that's all dkl
21:45dylanit last failed with "Build failed: The command '/bin/sh -c' returned a non-zero code: 1"
21:45dylandavidwalsh: know the github repo that backs it?
21:46dylanfound it
21:46dylanI don't have perms there, but someone that does could trigger a rebuild from that builds page
21:47dylanor commit something to
21:48dylanat some point you probably should use
21:54ckolosThe new phabricator interface tweaks are nice
22:07davidwalshmcote: smacleod: glob: Someone add me to this repo, please; need to merge
12 Aug 2017
No messages
Last message: 10 days and 22 hours ago