mozilla :: #screenshots

16 May 2017
14:31GitHub[screenshots] johngruen commented on issue #2844: @akhilman hey that's no good. So that's an older version of the add-on (we're moving to ship Page Shot natively in Firefox and the Test Pilot add-on is a few versions behind). The newer version is available in Mightly behind an about:config pref if you'd like to test it out (just search for a pref called "Screenshots").... https://git.io/v9bNM
14:31GitHub[screenshots] johngruen commented on issue #2844: @akhilman hey that's no good. So that's an older version of the add-on (we're moving to ship Page Shot natively in Firefox and the Test Pilot add-on is a few versions behind). The newer version is available in Nightly behind an about:config pref if you'd like to test it out (just search for a pref called "Screenshots").... https://git.io/v9bNM
16:26ianbickingugh, my local talos results not easy to acquire are useless
16:26ianbickingmake no sense, probably just noise
16:46GitHub[screenshots] 6a68 commented on issue #2799: Note that webextensions in a separate process is due to land in 55, so we should see any breakage in nightly and dev edition sooner than later https://git.io/v9NnY
16:49_6a68ianbicking: have you figured out how to run multiple runs in a row?
16:49_6a68I didn't see that in the mach talos-test options, I'm thinking maybe it's in mozharness somewhere
16:52ianbicking_6a68: it seems to do so automatically, maybe 10 times?
16:53_6a68ianbicking: oh, is that what the JSON file says? I thought it was running a bunch of different tests, that makes sense
16:53_6a68ianbicking: you were saying the numbers weren't meaningful?
16:53ianbicking_6a68: yeah, they arent that similar to the Try runs, and often are faster when they shouldnt be
16:54_6a68ianbicking: so I found this page yesterday that explained how to use the exact arguments from the Try runs when running tests locally, it seemed about half out of date
16:54_6a68ah yes, here it is: https://wiki.mozilla.org/ReleaseEngineering/Mozharness/How_to_run_tests_as_a_developer
16:55ianbickingI guess Ive lost faith, at least unless I close out everything and carefully do nothing on this computer while the tests run
16:55ianbickingoptions seem unlikely to sufficient to change things
16:56_6a68Have you tried a non-artifact build?
16:56ianbicking_6a68: considering how long the last build was, its probably non-artifact?
16:56_6a68It'll take 90 minutes, not 90 seconds, but the artifact build docs explicitly say they aren't known to work with Talos tests
16:56_6a68ah :-)
16:56_6a68I asked in #releng if the docs I linked are the best available, hopefully we'll get a response
16:56ianbicking_6a68: but all Ive done is edit a few files in the extension, done a small ./mach build, then run the talos tests
16:57ianbickingI dont distrust the build, I distrust the local talos tests
16:57ianbickingI didnt try to compare any runs that werent off almost the exact same Firefox build
16:58_6a68<jmaher> _6a68: the local numbers will not come close to try; in many cases you can do an A/B test with a patch locally and see the difference
16:58_6a68Supposedly ./mach talos-test -a testname should work
16:58ianbickingyeah; the A/B isnt working :(
16:58_6a68well, I got a linux machine set up yesterday, but didn&#39;t A/B anything. I can try that now, should only take a few minutes
16:58ianbickingIm doing A/B, but not seeing numbers that make sense
16:59_6a68care to pastebin the numbers? curious what you&#39;re seeing
16:59_6a68ianbicking: fwiw, setting up artifact builds is really easy, you just add a couple of lines to a .mozconfig file, let me dig up the link
16:59_6a68the builds actually finish in under 2 minutes for me with artifact build
17:00ianbicking_https://gist.github.com/ianb/55f91007251d1b4506d5c2f5027939e3
17:00_6a68here&#39;s the artifact build incantation: https://www.ncalexander.net/blog/2015/12/31/firefox-artifact-builds-for-mac/
17:00ianbickingthe t50n tests are probably not worth running, I think they were the slowest
17:01_6a68have you tried running the tests with screenshots removed entirely?
17:01_6a68still, pretty weird that some tests ran faster with screenshots on vs off
17:02ianbicking_6a68: no, only prefd off, but thats what weve been comparing
17:02_6a68right, I suspect that may not be the whole story, since we do still call LegacyExtensionUtils.getEmbeddedThing even if it&#39;s preffed off
17:02_6a68might be worth a try, at least
17:02_6a68I&#39;ll try the same
17:02ianbicking_6a68: I could, but it wont tell us anything about the performance concern we do have
17:03ianbickingmaybe over lunch
17:03_6a68:-\
17:03_6a68let me see if I can come up with data
17:03ianbickingI ran it between chores and stuff last night, otherwise I double-dont-trust the number if Im using the computer while the tests run
17:04_6a68yup
17:35_6a68ughhh, I hate working with hg
17:35_6a68ianbicking: are you just running tests against central?
17:35ianbicking_6a68: yeah
17:36_6a68I cannot get the right commit on the linux machine, hg2git is failing me
17:36_6a68but I was trying on beta, maybe central will work
17:36ianbicking_6a68: I guess beta would be better, since we understand what we should see there
17:57_6a68ianbicking: so, apparently there&#39;s a bug in beta where artifact builds fail
17:57_6a68this is probably why you went with the full local build approach
17:57_6a68I asked in fx-team and the fix is apparently to apply this one-liner: https://reviewboard.mozilla.org/r/133490/diff/3#index_header
18:00_6a68actually, that fix didn&#39;t work, never mind
18:02clouserw_6a68: standup
18:02_6a68
18:19GitHub[screenshots] ianb created bootstrap-button (+1 new commit): https://git.io/v9Nae
18:19GitHubscreenshots/bootstrap-button d27dba5 Ian Bicking: WIP moving button into bootstrap.js...
18:19ianbicking_6a68: ^^ that branch has my incomplete work
18:19_6a68sweet, thanks
18:20ianbicking_6a68: avoiding the stylesheet and jar.mn entirely might be easier than figuring them out, unless you already know how all that works
18:20_6a68oh, haha, jar.mn. sadly this stuff rings a bell
18:21GitHub[screenshots] wresuolc commented on issue #2844: I can&#39;t reproduce this on that domain in your screenshot. Can you try it with a clean profile (or just with fewer add-ons)? It might be conflicting with something else you have installed. https://git.io/v9NaY
18:21ianbicking_6a68: for getting the image and stylesheet into the Firefox build so they are URL-addressable :-/
18:21_6a68pretty sure I can nab some code out of the old universla-search addon to get the stylesheets and images working
18:28_6a68ianbicking: about the icons, did you try adding an entry in chrome.manifest? see the use of the &#39;skin&#39; directory here https://github.com/mozilla/universal-search/search?utf8=%E2%9C%93&q=skin&type=
18:28ianbicking_6a68: I didnt try that
18:29_6a68
18:29_6a68ianbicking: also, apparently the artifact build is failing for beta / linux because no artifacts were built on the mozilla servers
18:29_6a68so, I guess I&quot;ll crack open my ancient *windows* laptop and use that for running talos tests
18:32GitHub[screenshots] ianb created version-7.0.0 (+1 new commit): https://git.io/v9Nwe
18:32GitHubscreenshots/version-7.0.0 b35ed37 Ian Bicking: Update version to 7.0.0
18:32GitHub[screenshots] ianb opened pull request #2845: Update version to 7.0.0 (master...version-7.0.0) https://git.io/v9NwJ
18:33GitHub[screenshots] ianb pushed 1 new commit to master: https://git.io/v9NwT
18:33GitHubscreenshots/master 17ae5e4 Ian Bicking: Merge pull request #2845 from mozilla-services/version-7.0.0...
18:33GitHub[screenshots] ianb deleted version-7.0.0 at b35ed37: https://git.io/v9Nwk
18:33GitHub[screenshots] ianb merged master into stable: https://git.io/v9Nwq
18:34GitHub[screenshots] johngruen opened issue #2846: Let&#39;s add UTM parameters for various places where people can copy shot URLS https://git.io/v9Nwc
18:36jgruenhttps://github.com/mozilla-services/screenshots/issues/2846
18:36jgruenclouserw: ^
18:38cloudops-ansiblescreenshots build #36: building mozilla/pageshot:7.0.0
18:39clouserwjgruen: I can sync up w/ gareth on it
18:39jgruensending email now
18:39clouserwk
18:57GitHub[screenshots] 6a68 created bootstrap-button-wip at 06dbaab (+0 new commits): https://git.io/v9NKj
19:03cloudops-ansiblescreenshots build #36: mozilla/pageshot:7.0.0 deployed to Stage
19:18cloudops-ansiblescreenshots build #36: mozilla/pageshot:7.0.0 ready for prod
19:30GitHub[screenshots] dannycoates closed issue #2521: Setup continuous deployment https://git.io/vSs2Z
19:41cloudops-ansiblescreenshots-dev build #19: building mozilla/screenshots:latest
19:44cloudops-ansiblescreenshots-dev build #19: mozilla/screenshots:latest deployed to Dev
20:02jgruenmahe, clouserw, we meeting?
20:03jgruenoops, looking at the wrong day in my calendar
20:03mahejgruen.. Cool..was wondering I was looking at the wrong day
20:03jgruennope, that&#39;s on me :)
20:03jgruensee you tomorrow!
20:03mahesee you!
20:09GitHub[screenshots] ianb created lazy-background (+1 new commit): https://git.io/v9NQY
20:09GitHubscreenshots/lazy-background 0009e12 Ian Bicking: Fix #2843, lazily load code into the background page...
20:09GitHub[screenshots] ianb opened pull request #2847: Fix #2843, lazily load code into the background page (master...lazy-background) https://git.io/v9NQZ
20:27GitHub[screenshots] ianb force-pushed lazy-background from 0009e12 to 12e2bf9: https://git.io/v9N51
20:27GitHubscreenshots/lazy-background 12e2bf9 Ian Bicking: Fix #2843, lazily load code into the background page...
20:34circleci-botFixed: ianb&#39;s build (#1429; push) in mozilla-services/screenshots (lazy-background) -- https://circleci.com/gh/mozilla-services/screenshots/1429?utm_campaign=chatroom-integration&amp;utm_medium=referral&amp;utm_source=irc
21:05GitHub[screenshots] garethcull commented on issue #2846: Sure thing John. Comments below for each. Append the following to each url:... https://git.io/v9Nx7
23:31_6a68There must be a Mozilla drinking game for each time you wait an hour, then discover you&#39;ve hit an undocumented, known bug
17 May 2017
No messages
   
Last message: 11 days and 2 hours ago