mozilla :: #ateam

18 May 2017
07:43AutomatedTesterwhere are the mozreview docs for how to revive a reviewboard autoland for backouts
09:22davehuntglob: just read your blog post, I think I need to draw my own diagram :) what tool did you use?
10:55atoAutomatedTester: Is this change in a public release? https://github.com/SeleniumHQ/selenium/commit/01ac0cce04ae4607ca3dc63f7df75fce2e183839
11:04whimbooato: hello
11:04whimboothank you for the email re window handling
11:04whimboodoes it mean that you want to get started on it?
11:54atowhimboo: Ive started coding a prototype, but Im not sure what bug the work Im doing will end up belonging to.
11:55atowhimboo: I know its not going to fix the fallout from Daos change, so I think we need to work on that separately.
11:55atowhimboo: But what Im working on will fix the remoteness bug (https://bugzilla.mozilla.org/show_bug.cgi?id=1311041).
11:55bugbotBug 1311041: Marionette, normal, ato, ASSIGNED , Window handle changes on remoteness change
11:56whimbooato: i fear to start with something which you might have changed
11:56whimboochanging the issue as introduced by dao will need a rewrite
11:57atowhimboo: The fallout seems to be about waiting for the right set of conditions when creating a new session.
11:57whimbooits hard to simply patch this single case
11:57atoReally? That wasnt my impression, but maybe I dont have the full grasp of the issue.
11:57atoI mean, sure, there are confounding problems but waiting for the session restoration to finish before returning from the newSession command doesnt sound impossible with the current infrastructure?
11:58whimbooato: we could only fix this particular issue on startup
12:15jgrahamwhimboo: Everything's in the patchset
12:15whimboojgraham: sorry, I don't have time to read everything. so i would appreciate some condensed information
12:18jgrahamTypically I can't work out how to link to anything in mozreview so look at https://reviewboard.mozilla.org/r/137974/diff/1#index_header testing/marionette/reftest.js line 65
12:23whimboojgraham: would you mind adding this to the comment? I cannot look at this right now
12:23jgrahamSure
12:24whimboothanks
12:27jgrahammaja_zf: https://github.com/jgraham/mozlog_rust is something i did on the train, in case we need it
12:28jgrahamSame is easy to do in Python, but this could be faster &c.
12:28jgrahamDunno if it will be useful, but it was more fun than staring out of the window
12:32Tomcat|sheriffdutyjmaher|afk: ping
12:37AutomatedTesterato: it should be
12:37AutomatedTesterato: hmmm might not be
12:39atoAutomatedTester: Someone wrote to selenium-users@ about the bug you appear to have fixed, but its not easy to track version information of the Python package from the git history.
12:39atoOr maybe it is and I dont understand how to (-:
12:39AutomatedTesterato: looking at the changelog its not documented
12:39AutomatedTesterso I am guessing I havent released it
12:40AutomatedTesterato: no... see https://github.com/SeleniumHQ/selenium/commits/master/py
12:40AutomatedTesterI can do a release now
12:41atoRight, thanks for checking!
12:41AutomatedTesterato: click the history for the directory
12:53whimboojgraham: map() should directly work on Result<>...
12:53whimboohttps://github.com/jgraham/mozlog_rust/blob/master/src/main.rs#L25
12:55jgrahamwhimboo: It&#39;s Result<Vec<T>> or something&#39; I
12:55jgraham; I&#39;m trying to map the Vec not the Result
12:55whimboocorrect
12:55whimboothere is .map and .map_err
12:55whimboo.map will work on the value of Result
12:56jgrahamYeah, but it would do something different
12:57jgrahamIf I have Ok(vec![1]) and I call .map(|x| x) on it I end up with x == vec![1]. If I call .unwrap().map(|x| x) on it, I end up with x == 1
12:57AutomatedTesterato: new Selenium python bindings released
12:57atoAutomatedTester: Yay!
12:57atoAutomatedTester: Do you want to reply to the thread on selenium-users@ or shall I?
12:58AutomatedTesterato: I will let you :D
12:58jgraham(also this is Option<> not Result<>, although the same applies)
12:58atoOK, will do it later.
12:58atoIn my next email batch.
12:58whimboojgraham: hm k. at least paths are required to unwrap() doens&#39;t cause problems
12:58jgrahamIn that case the .wrap() is expected to be safe beecause clap guaranteeds that an argument with required(true) is Some(x), not None
12:59jgrahamI could .expect() or something
13:00jgrahamOr handle it gracefully with the same result (the program exits if None)
13:00jgrahamBut train code :)
13:00whimbooheh
13:01whimboojgraham: my flight yesterday from Paris was kinda bumpy
13:01jgrahamUgh
13:01whimboonothing I&#39;m in favor with
13:01jgrahamI was very tired on the way home; it wasn&#39;t fun
13:02whimbooi feel with you
13:02whimboothose three days were exhausting
13:03whimbooato: interesting ux for marionette! i like it
13:04whimboobut the left side of the urlbar looks strange. somewhat cut off
13:04atowhimboo: Best I could do with the time given to me.
13:04atoBut the basic plumbing to make it happen is now in place.
13:04whimbooyeah, no problem
13:04atoSo we can iterate from here.
13:04whimboojust cosmetics
13:05atoRight. The hard part about this patch wasnt the styling but getting the Marionette-to-<xul:browser> communication right.
13:05whimbooright
13:27globdavehunt: i see you found a tool :) i use shapes (shapesapp.com) but any simple tool should work
13:27davehuntglob: I realised I had an old version of OmniGraffle installed... you can see my image is heavily inspired by yours :P
14:20sewardjted: greetings.
14:21sewardjted: we used to have an extension called &quot;quitter&quot;. Do you know if that still exists?
14:40jmahersewardj: I believe it still exists
14:42kumar_Hi, I am new here, kindly guide me to contribute. Where can I start?
14:43jmahersewardj: https://dxr.mozilla.org/mozilla-central/source/tools/quitter
14:43jmaherTomcat|sheriffduty: pong
14:46jmaherAutomatedTester: do you know if when using Marionette it changes the profile/state of the browser for a future session? I have 2 specific talos tests that are failing only when using marionette to close the browser while warming up the profile before running the test
14:46jmaherall the other talos tests pass just fine; but two which share the same addon are failing
14:47sewardjjmaher: thank you
14:48Tomcat|sheriffdutyjmaher: unping all done :))
14:49jmaherTomcat|sheriffduty: ok
14:52kumar_jmaher: Can you help me please :)
14:53szlkumar_
14:55kumar_szl: ??
14:55szlSorry that was unintentional
14:56jmaherkumar_: what do you need help with?
14:57kumar_Can you link me to contribution/dev setup guides?
14:58jmaherkumar_: what type of technologies do you want to work with? or what languages are you most familiar with or want to learn?
14:59kumar_I am familiar with Python and C++ and willing to learn other languages also, need a starting point. :)
15:00jmaherkumar_: what interests you about contributing to Mozilla?
15:00jmaherthere are so many places to start, I don&#39;t want to ask you to help on something that is not interesting
15:02kumar_I am new to open source projects and I guess starting with Mozilla is the a good thing . Can I find a list of projects and the languages involved to be able to find something interesting.
15:04kumarHi kumar_, high five for having a cool name. If you&#39;re interested in Python we do a lot of it in Mozilla&#39;s websites. You can find out about getting involved here: https://wiki.mozilla.org/Webdev/GetInvolved Some projects on that list are no longer active so just check the repository activity first.
15:05jmaherkumar_: do you like test automation, web development, hacking on firefox backend, firefox frontend, other services, tools ?
15:05jmaherthe ateam is about test automation
15:06jmaherkumar_: for test automation here is an outdated list of projects: https://wiki.mozilla.org/index.php?title=EngineeringProductivity/Projects
15:07jmaherabout 6 months old, but many things are still relevant (links to bugzilla searches, etc.)
15:15kumar_I have decided to try out Talos . Thanks a lot kumar and jmaher :)
15:16jmaherkumar_: cool, rwood and myself are working on talos quite actively right now
15:16kumar_https://wiki.mozilla.org/Buildbot/Talos/Running#Running_locally_-_Source_Code Is this the right place to start?
15:17jmaherkumar_: yes, that would be a good place to start, ensure you can run it locally
15:33atowhimboo|bbl: What do I need cause a remoteness change?
15:34whimbooato: navigate to about:robots and then to any http/https address
15:34whimbooand vice versa
15:34atowhimboo: Thanks!
15:35whimbooato: see my test in test_click.py
15:35whimbootst_click_remoteness_change
15:35whimboo+e
15:35jmaherAutomatedTester: hmm, running on a loaner, it looks like Firefox is crashing on tart/cart tests, not necessarily timing out
15:35jmaherthis is only when using marionette to warm up the profile :(
16:33atowhimboo, AutomatedTester: Status update on my browser tracking prototype is that it appears to work flawlessly with remoteness changes, e.g. the window handle and element store is preserved.
16:37whimbooato: oh, cool!
16:37whimbooato: how many changes you had to do?
16:37atowhimboo: Ive only coded a replacement that watches browsers appear and disappear for now, so I havent integrated it with the rest of Marionette.
16:38atoBut the exploratory tests I have show that we keep the UUID I assign to the browser after a remoteness change.
16:38atoAlso it appears pretty non-invasive to the rest of Marionette, which is good.
16:39whimboonice
16:39atoI think I will associate this patch with the remoteness bug Im already assigned to.
16:39whimboosure
16:39whimbooato: i&#39;m on click + window close now
16:39atoBut still plenty of work left to do. I estimate another two or three weeks of work.
16:39whimbooso i might wait until you finished yours
16:39whimbooups
16:39whimbook, then not :D
16:40atoIm being generous because I havent run this on Fennec yet.
16:40atoMight of course take less time than that with the right focus.
16:40atoToday has been pretty good.
16:41jgrahamSpeak for yourself :p
16:41jgraham(but seriously that sounds great)
16:41atoAnd I think its possible to land it in stages. I have some ideas about replacing the broadcast message manager ugliness we use with per-browser message channels.
16:42atoFor every message we send to listener.js, we currently broadcast it to all browsers with the message name containing a unique number.
16:42atoThen each listener looks at the message and only the listener with the right number answers.
16:43atoThere are possible improvements to be made to that model.
16:44atoNow that we have a map of UUID browser, we can just open a channel directly to that browser.
16:44atoBut that would involve a tonne of changes.
16:44atoAlso it will be nice not to keep growing our memory usage whenever a remoteness change invalidates the reference to the weak refs we have to DOM elements.
16:44jgrahamI think today would improve if I didn&#39;t have The Gruffalo Child song in my head
16:45atoAnd it paves the way for serious changes to browser.js, which I think we ought to burn with fire.
16:45atoHah.
16:48tedsewardj: sorry, i was out for a run then eating lunch etc, still around?
16:49whimboojgraham: heh. my daughter currently reads this book
16:52atogps: Do you think youll have a spare moment to look at https://bugzilla.mozilla.org/show_bug.cgi?id=1340637 today?
16:52whimbook, i think I&#39;m done for today
16:52bugbotBug 1340637: Marionette, normal, ato, ASSIGNED , Import and build geckodriver in central
16:52whimboosee you tomorrow
16:52atowhimboo: Bye now. Im just leaving too.
16:53jgrahamwhimboo: In that case I either highly recommend, or don&#39;t recommend at all, getting the CD of songs, depending for your tolerance for havivng the tunes stuck in your head all day. Although I guess they probably aren&#39;t in German so maybe you don&#39;t have to make that choice
18:33whimboojgraham: so topic of school today.... The fox!
18:33whimboos/of/in
18:38Silne30whimboo: Are the update tests still having failures?
18:38whimbooi don&#39;t know yet
18:39whimboothe necessary patch hasn&#39;t been backported yet to unbreak update tests
18:46Silne30OK. To run ./mach firefox-ui-update, do I need to build firefox first?
18:46Silne30Or is that part of the whole mach process.
18:57whimbooyou need --binary
18:57whimbooand have to use an older release of firefox
18:57whimbooso grab the 2nd last beta release
20:44jgrahamjames@ginny:~/develop/gecko$ ./mach wpt --product chrome testing/web-platform/tests/dom/historical.html
20:45jgraham[]
20:45jgrahamRan 72 tests (1 parents, 71 subtests)
20:47AutomatedTesterjgraham: you just ran Chromium via mach?
21:25jgrahamAutomatedTester: Yes
21:46erahmwlach: is it possible mozregression is broken for date-based bisection?
21:56erahmoh ffs, it&#39;s probably because I want debug builds
21:59wlacherahm: hmm, I figure that of all things should work
22:00erahmwlach: I had my repo set to m-i and was specifying debug, I figure it&#39;s one of those two
22:01wlachhm
22:01erahmwlach: I&#39;ll see if I can repro w/ the command line tool
22:05erahmwlach: hmm, I think for debug it defaults to inbound but does some sort of date => cset mapping, I should probably copy that logic I guess. Would be nice if get_build_info or whatever it&#39;s called handled that
19 May 2017
No messages
   
Last message: 92 days and 19 hours ago