[08:44:29] <joakino>	 morning
[08:54:11] <joakino>	 hashar: are you around? i have a question about the filesystem structure of the sources in CI
[09:00:33] <hashar>	 joakino: hello
[09:00:41] <hashar>	 joakino: what do you want to know?
[09:00:42] <joakino>	 hey!
[09:01:03] <joakino>	 hashar: i'm working on a command line tool that I run on npm test on patches
[09:01:08] <joakino>	 sample patch here https://gerrit.wikimedia.org/r/#/c/312850/
[09:01:27] <hashar>	 eeeeeeee :D
[09:01:47] <joakino>	 hashar: the CLI needs to read the sources from the extension it is configured in, and from core too, so I was assuming you can read ../../resources to get core's frontend assets, but it seems to fail
[09:02:11] <hashar>	 ah so kind of an integration test between MF and MediaWiki
[09:02:17] <hashar>	 but mw/core is not cloned in the npm jobs
[09:02:21] <joakino>	 oooh
[09:02:39] <joakino>	 i thought all the sources where there
[09:02:41] <hashar>	 the mediawiki-extensions-hhvm  does though
[09:02:47] <hashar>	 but it runs the PHPUnit suite from MediaWiki core
[09:03:07] <joakino>	 darn haha
[09:03:15] <hashar>	 what is that resource-modules lint thing ?
[09:03:37] <hashar>	 in MediaWiki there are a few suite of tests that are shared/run for all extensions
[09:03:45] <joakino>	 it analyzes JS sources from an extension and core and checks the dependency graph against resourcemodules configuration in extension.json
[09:03:46] <hashar>	 including one that test ResourceLoader material
[09:04:37] <joakino>	 hashar: this is more a per-extension linter, but needs to read core to check JS definitions
[09:04:46] <joakino>	 that's why i thought of running it in npm test
[09:05:11] * hashar looks at https://www.npmjs.com/package/resource-modules
[09:05:17] <joakino>	 yep
[09:05:58] <joakino>	 would it be possible to clone core's sources on the npm job? it doesn't need to run mediawiki or anything
[09:06:22] <joakino>	 it's an experimental thing, I'm trying to get it set up to start polishing it
[09:08:12] <hashar>	 I am not sure what resource-modules is doing really
[09:09:17] <joakino>	 hashar: i didn't explain it very well hah
[09:09:28] <hashar>	 at least the npm-node-4  runs on disposable instances, and we can clone mw/core quite fast on those instances
[09:09:48] <hashar>	 then the job is triggered by any repo, most would not need mw/core to be cloned. But we can craft yet another job
[09:10:35] <joakino>	 hashar: maybe fetching a snapshot of master would be faster instead of cloning a repo with the history?
[09:11:19] <hashar>	 but from what I am reading, that looks to overlap with the PHPUnit tests/phpunit/structure/ResourcesTest.php and or ExtensionJsonValidationTest.php
[09:11:31] <hashar>	 ah
[09:11:47] <hashar>	 on disposable instances  mediawiki/core is already included as a mirror (something like /srv/git/mediawiki/core.git )
[09:12:18] <hashar>	 so in the workspace the job clones using that mirror as a reference.  They are on the same filesystem so git clone just hardlink which is a few miliseconds :D
[09:12:26] <hashar>	 so essentially it is not a problem
[09:12:28] <joakino>	 hashar: it overlaps a bit with that, but this cli tool reads the sources, not just the extensionljson
[09:12:43] <joakino>	 awesome
[09:13:22] <joakino>	 so you're saying to make a different npm job that clones core too to not affect all extensions that use the normal npm job
[09:13:47] <hashar>	 <rant>oh man I hate javascript, looks like a soup of random characters :D</rant>
[09:13:53] <hashar>	 yeah
[09:13:57] <hashar>	 or not I am not sure
[09:14:17] <joakino>	 lol
[09:15:21] <hashar>	 hmm
[09:15:30] <hashar>	 we will need another job I think
[09:15:47] <hashar>	 cause surely you will end up needing a whole lot other extensions to be added at some point
[09:15:51] <hashar>	 and probably skin as well
[09:16:35] <joakino>	 hashar: the cli only assumes existence of core sources, nothing else
[09:16:55] <joakino>	 i don't think frontend assets depend on frontend assets of other extensions
[09:16:58] <hashar>	 yeah until you get core that defines material depending on Vector
[09:17:09] <hashar>	 or MF depending somehow on VisualEditor or whatever else :]
[09:17:21] <hashar>	 regardless, creating a new job is fairly trivial
[09:17:51] <joakino>	 yeah true
[09:18:06] <joakino>	 another question is if that is *lintable* haha
[09:18:26] <hashar>	 and we can make it to run npm install && npm run-script lint:modules
[09:18:28] <hashar>	 or something like that
[09:18:51] <joakino>	 hashar: i'd love a job what run npm run ci to do different things on that env
[09:19:32] <joakino>	 or npm run test:ci
[09:20:27] <hashar>	 ReferenceError: Promise is not defined
[09:20:33] <hashar>	 ahoh ou my nodejs must be outdated :D
[09:20:37] <joakino>	 haha
[09:21:15] <joakino>	 needs 4+ like the ci environment
[09:21:28] <hashar>	 ah
[09:21:32] <hashar>	 yeah that would help I guess
[09:21:39] * hashar shakes fists at Debian outdated packages
[09:22:32] <hashar>	 and obviously upgrading nodejs ends up uninstalling python and kde
[09:22:34] <hashar>	 :D
[09:23:35] <joakino>	 classic nodejs
[09:23:53] <hashar>	 anyway
[09:24:03] <hashar>	 your linting tool does interpret javascript
[09:24:22] <hashar>	 and it is different than the MediaWiki core PHPUnit test suites that look at RL definition is that correct?
[09:25:03] <joakino>	 yep
[09:25:15] <joakino>	 checks that what the source says/uses is reflected properly on extensionjson
[09:25:47] <joakino>	 the mw php RL linting checks that extension.json is correct (cyclic graphs, etc)
[09:25:58] <joakino>	 (i believe)
[09:26:16] <hashar>	 so my point being
[09:26:36] <hashar>	 if stuff can be added to the mediawiki/core PHPUnit  suite, that will benefit all extensions
[09:26:39] <hashar>	 BUT
[09:27:04] <hashar>	 some checks/lint would need to interpret javascript and thus would need a dedicated job (which is fine to me :D )
[09:27:26] <joakino>	 definitely, that'd be lovely. it needs a lot of polishing first though
[09:27:33] <hashar>	 I am not even sure what  tests/phpunit/structure/ResourcesTest.php does in core
[09:27:51] <hashar>	 but it seems to warn about missing messages and dependencies at least
[09:29:23] <joakino>	 hashar: from source i see illegal dependencies, missing dependencies, missing messages, unsatisfiable deps (targets related)
[09:29:54] <joakino>	 all ResourceModules config related, it is good
[09:33:45] <hashar>	 is that from the PHPUnit suite ?
[09:36:51] <joakino>	 from tests/phpunit/structure/ResourcesTest.php
[09:38:58] <hashar>	 meanwhile I am crafting a job :D
[09:39:11] <joakino>	 thanks hashar!
[09:40:41] <hashar>	 one sure thing, is some tests can be embeded in the PHPUnit suite, that would benefit everyone with little to no work required :]
[09:40:52] <hashar>	 once the test lands in mediawiki, all extensions get it run for them
[09:42:30] <joakino>	 that's a good endgoal
[09:42:59] <joakino>	 parsing source and JS code dependencies is non-trivial and i'm not sure to what extent it can be generalized
[09:43:12] <joakino>	 that's why i want to test with a few extensions first
[09:43:16] <hashar>	 yup
[09:46:48] <hashar>	 and of course Jenkins does not support ':' in a job name (eg: lint:modules) bah
[09:55:01] * hashar fights with yaml
[09:58:00] <joakino>	 hashar: want company?
[10:01:32] <hashar>	 I am trying something dirty :D
[10:02:00] <hashar>	 gotta find some time to refactor the crap
[10:02:07] <hashar>	 and drop a lot of duplication we have :(
[10:03:03] <joakino>	 hashar: btw you this cli works globally too so if you npm install -g resource-modules && resource-modules ./extension-folder that should work too
[10:03:36] <joakino>	 hashar: the yml config definitely needs love
[10:07:55] <hashar>	 well we are in a middle of crossing a bridge
[10:08:01] <hashar>	 with bunch of jobs being duplicated :(((
[10:08:43] <joakino>	 :(
[10:13:17] <hashar>	 joakino: giving a try at https://integration.wikimedia.org/ci/job/mwext-MobileFrontend-npm-run-lint-modules/1/console
[10:13:27] <hashar>	 and obviously it explodes completely :]
[10:14:02] <joakino>	 haha
[10:14:04] <hashar>	 but yeah the idea is to clone mw+MF
[10:14:19] <hashar>	 under /src
[10:14:23] <hashar>	 err
[10:14:26] <hashar>	 under "$WORKSPACE/src
[10:14:40] <hashar>	 then  cd src ; npm install && npm run lint:modules
[10:15:04] <hashar>	 the template I came with is not reusable / hardcoded for MF
[10:15:09] <hashar>	 but that is probably good enough for now
[10:15:17] <joakino>	 definitely hashar 
[10:15:33] <joakino>	 so it'll be src/mediawiki & src/MobileFrontend?
[10:15:39] <hashar>	 https://integration.wikimedia.org/ci/job/mwext-MobileFrontend-npm-run-lint-modules/2/console
[10:15:42] <hashar>	 stuff happening maybe
[10:15:45] <hashar>	 no
[10:15:51] <hashar>	 mediawiki core will be in $WORKSPACE/src
[10:16:02] <hashar>	 and the extension in  $WORKSPACE/src/extensions/MobileFrontend
[10:16:11] <hashar>	 so in theory, your relative links might work just fine
[10:16:41] <joakino>	 not really, since extensions is usually inside mediawiki/
[10:16:52] <joakino>	 but that's fine, i'll fix the cli to take another argument to core
[10:17:42] <hashar>	 you mean lint:modules ?
[10:17:45] <hashar>	 it is fine :]
[10:19:13] <hashar>	 one more time this time with a cd src/extensions/MobileFrontend  :D
[10:19:13] <hashar>	 https://integration.wikimedia.org/ci/job/mwext-MobileFrontend-npm-run-lint-modules/3/console
[10:20:22] <hashar>	 joakino: ^^  seems it fails gracefully :D
[10:21:05] <joakino>	 hashar: that is actually the correct failures!
[10:21:17] <joakino>	 \o/
[10:21:37] <joakino>	 oh i get it now, core is on src/
[10:21:40] <joakino>	 you're right
[10:21:51] <hashar>	 yeah
[10:21:59] <hashar>	 so that is a bit confusing
[10:22:09] <hashar>	 but some jobs clone the repositories under /src/
[10:22:21] <hashar>	 so we can add other directories that do not conflict with the source repos
[10:22:38] <hashar>	 such as /log/   (where one can land logs / artifacts which are then attached to the build)
[10:22:40] <hashar>	 or /caches/
[10:22:43] <hashar>	 or whatever else
[10:22:49] <joakino>	 aah, makes sense
[10:23:12] <hashar>	 but most old jobs are cloning directly at the root of the workspace
[10:23:17] <hashar>	 so that is very confusing :(
[10:23:32] <joakino>	 indeed
[10:24:38] <joakino>	 hashar: so if we add this job to mobilefrontend, i'll have to submit a patch with "lint:modules" empty for patches to pass, right?
[10:24:50] <joakino>	 until https://gerrit.wikimedia.org/r/#/c/312850/ is ready to merge
[10:26:18] <hashar>	 nop
[10:26:37] <hashar>	 I am going to add it to the 'experimental' pipeline
[10:26:43] <hashar>	 so patches do not trigger the job
[10:26:52] <joakino>	 aha
[10:26:53] <hashar>	 and on your change, you will have to comment 'check experimental' to run it
[10:27:01] <joakino>	 awesome
[10:27:04] <joakino>	 !
[10:27:23] <joakino>	 gonna remove lint:modules from npm test on that patch then
[10:27:39] <hashar>	 do you have a phabricator about it ?
[10:27:45] <hashar>	 so I can attach the CI config change to the task
[10:27:50] <hashar>	 if there are no task, lets skip :]
[10:28:22] <joakino>	 hashar: there's no task right now
[10:28:26] <joakino>	 should i create one?
[10:30:54] <joakino>	 i'm on it
[10:31:45] <hashar>	 joakino: up to you :]
[10:32:01] <joakino>	 hashar: use T146748
[10:32:03] <stashbot>	 T146748: Lint JS sources dependency graph against ResourceModules configuration - https://phabricator.wikimedia.org/T146748
[10:32:05] <joakino>	 i'll fill it up in a moment
[10:35:17] <hashar>	 I am deploying the job and CI Config
[10:35:57] <hashar>	 during Releng meeting yesterday we were wondering how much devs care about testing
[10:36:08] <hashar>	 and I think we are gonna take mobile as an example :D
[10:39:23] <hashar>	 I have commented "check experimental" in Gerrit and some stuff is going to explode at https://integration.wikimedia.org/ci/job/mwext-MobileFrontend-npm-run-lint-modules/5/console
[10:41:17] <hashar>	 joakino: looks "good" to me. Let me know if something has to be polished up
[10:42:20] <hashar>	 if you have the job to craft a report / a junit file.  You can it get it written to $WORKSPACE/log   (which might be  relative dir  ../../../log
[10:42:37] <joakino>	 that looks great hashar, thanks a lot
[10:42:43] <joakino>	 i owe you a beer
[10:42:44] <hashar>	 and the build system will capture the log and attach it to the build.  If its a junit file we can enable the jenkins plugin that craft a report for it
[10:42:50] <hashar>	 yeah yeah beer :D
[10:43:27] <hashar>	 and potentially
[10:43:44] <hashar>	 we can get that added to mediawiki/core as well
[10:44:29] <joakino>	 awesome
[10:44:38] <joakino>	 i'll look into that once it is more mature
[10:46:02] <hashar>	 yup
[10:46:14] <hashar>	 and add colors ! :D
[10:46:28] <hashar>	 for some reason devs pay attention to anything written in red
[10:47:50] <joakino>	 hashar: colors on the terminal output appear on https://integration.wikimedia.org/ci/job/mwext-MobileFrontend-npm-run-lint-modules/5/console ?
[10:48:08] <hashar>	 though one might argue that in latin american green is associated with death  and in Asia red is happiness :D
[10:48:15] <hashar>	 need some geolocalization to alter the colors
[10:48:22] <hashar>	 yeah
[10:48:28] <hashar>	 so if you spurt ansi term sequences
[10:48:38] <hashar>	 there should be a plugin that transform them to nice coloring in the console
[10:48:48] <hashar>	 though that is apparently broken for npm :/
[10:56:39] <joakino>	 oh :(
[10:56:51] <joakino>	 i'll use orange when I get to it
[11:48:08] <phuedx>	 ^
[11:48:11] <phuedx>	 AWESOME
[11:48:16] <phuedx>	 this whole exchange
[13:23:46] <_hc>	 hey all, I'm starting in on a merge request to add Tor support to the Android app.  It is currently failing to find support-v4.jar (com.android.support:support-v4:24.1.1), looking at $ANDROID_HOME/extras/android/m2repository/com/android/support/support-v4/24.1.1/ shows me only an aar, not a jar
[13:24:11] <_hc>	 any ideas here? (fyi, I've been doing Android since 2009, so this isn't a newbie question ;-)
[13:24:24] <_hc>	 I've never seen that error before
[13:25:39] <_hc>	 hmm, looks like this line was omitted from build.gradle:
[13:25:39] <_hc>	 compile 'com.android.support:support-v4:24.2.1'
[13:27:50] <_hc>	 nope that didn't do it
[13:27:59] <_hc>	 but bumping supportVersion = '24.2.1' got it working for me
[13:28:11] <_hc>	 perhaps 24.1.1 is the version where Google removed some binaries
[13:34:17] <mdholloway>	 _hc: o/ yeah, that error didn't ring a bell to me either and i was doing some digging around our gradle config, glad the supportVersion bump fixed it. 
[13:47:13] <phuedx>	 mdholloway, _hc: might be worth filing a bug in phabricator just so that the fix is archived and searchable!
[13:49:22] <_hc>	 I'm hoping to do a merge request soon, I can just include that
[13:51:05] <mdholloway>	 _hc: the patch sounds interesting, looking forward to it!
[13:51:16] * mdholloway looks to see if we have any outstanding tasks having to do with tor support
[13:51:17] <_hc>	 cool :)
[13:52:30] <_hc>	 while you're here, can you point me to where preferences are handled?  I'll add a "Use Tor" switch there, that's how its usually done
[13:52:56] <_hc>	 ok, preferences.xml with some extras
[13:54:10] <mdholloway>	 _hc: yes, the main user-oriented preferences are in preferences.xml, and we also have a fairly extensive set of developer prefs in developer_preferences.xml (as you probably saw)
[13:55:28] <_hc>	 I figure Privacy is the right secrion for "Use Tor".  Any suggestions on placement in the list of 2 items there?
[13:56:27] <mdholloway>	 _hc: yeah, the privacy section makes sense.  wherever you like seems fine
[13:56:40] <mdholloway>	 _hc: easy enough to change if someone else has strong feelings about it
[15:35:01] <dr0ptp4kt>	 phuedx: yt? ok if I resched?
[15:37:20] <dr0ptp4kt>	 phuedx: I'm gonna hop on another call. Will resched.
[15:56:23] <phuedx>	 dr0ptp4kt: darn -- sorry!
[15:56:34] <phuedx>	 my youngest son has what my oldest son had
[15:56:40] <phuedx>	 there was vomit all over the lounge
[15:56:44] <phuedx>	 shoud've marked myself as away
[16:02:23] <maxbinder>	 joakino: jdlrobson dr0ptp4kt nzr retro
[17:34:52] <niedzielski>	 nzr: o/ hello! hey, i'm blocking a bit on design feedback for T146726. please follow up when you can
[17:51:01] <niedzielski>	 dbrant nzr: tapping on a notification from google's stock calendar app is another example of the cross to close in the upper left design
[17:51:21] <dbrant>	 nice
[19:07:15] <mdholloway_>	 niedzielski-afk: bearND bearND_ (i guess dbrant isn't around): i'm having a look at the autohide patch now as well, works well, just about to take a look at the code
[19:49:00] <niedzielski>	 mdholloway_: thanks :)
[21:24:10] <nzr>	 niedzielski: can't open the alpha from the email 
[21:25:29] <niedzielski>	 nzr: ah, fixed
[21:25:57] <nzr>	 niedzielski: thanks
[21:26:18] <niedzielski>	 👍
[21:33:32] <nzr>	 niedzielski: same link should work?
[21:33:52] <niedzielski>	 nzr: i think so. i just fixed the privacy settings. i didn't realize files defaulted to being private
[21:34:24] <nzr>	 niedzielski: now it's asking me for phab password
[21:34:39] <nzr>	 can't download apk :( dont have my phab and wiki credentials on test device
[21:35:06] <niedzielski>	 nzr: works in an incognito window for me :|
[21:35:30] <niedzielski>	 nzr: i'll just email it directly
[21:36:10] <nzr>	 niedzielski: that would be wondeful :)
[21:42:52] <nzr>	 niedzielski: one minor regression. the vertical alignment for empty pages is not middle anymore. not a blocker but we should track
[21:43:52] <niedzielski>	 nzr: please open a phab with repro steps :]
[21:48:54] <nzr>	 niedzielski: yep
[22:03:07] <maxbinder>	 jdlrobson: just wanted to acknowledge receipt of emails, since I might not have a chance to address them until tomorrow (slammed by meetings over 4 teams :)
[22:17:06] <Volker_E>	 jdlrobson: this has to happen first: https://gerrit.wikimedia.org/r/#/c/313116/
[22:17:24] <Volker_E>	 jdlrobson: dependency for buttons and Minerva greys
[22:43:19] <Volker_E>	 jdlrobson: https://gerrit.wikimedia.org/r/#/c/313134/
[22:47:19] <Volker_E>	 jdlrobson: https://gerrit.wikimedia.org/r/#/c/313116/
[23:14:48] <nzr>	 joakino: o/ wanna take a look at print styles
[23:27:12] <Volker_E>	 jdlrobson: https://gerrit.wikimedia.org/r/#/c/312651/
[23:27:19] <Volker_E>	 jdlrobson: :)
[23:29:26] <jdlrobson>	 nzr: you know joaquin is long asleep right ? :) well he should be... or did you mean me?
[23:34:02] <nzr>	 jdlrobson: haha no.. i meant him but i guess he will see it whenever he gets up from his sweet dreams
[23:35:47] <jdlrobson>	 nzr: https://phabricator.wikimedia.org/T146839
[23:35:53] <jdlrobson>	 please comment/edit/close :)
[23:38:27] <nzr>	 jdlrobson: let's keep it in future. removing sprint?
[23:38:53] <jdlrobson>	 i dunno.. we want to push the new footer out this sprint, so i wasnt sure if it was a blocker
[23:39:00] <jdlrobson>	 if it's not a blocker remove the sprint and comment so