mozilla :: #ateam

6 Sep 2017
11:13whimbooato: i tried to make the changes to the async code for addon installs. whatever I do it returns too early now, and I have no f&*#$%g idea why :S
11:14atowhimboo: Post a .patch somewhere and I can take a look.
11:14atoLunch first, however.
11:14whimbooenjoy that first
11:14whimboomaybe i can get something working by then
11:57atohttps://blog.rust-lang.org/2017/09/05/Rust-2017-Survey-Results.html is interesting.
12:03whimbooato: got it resolved1
12:03whimboowill upload the new patch which should make you happier
12:06atoYay!
12:09whimbooato: thing is returning AddonManager.getInstallForFile(file).then(...) doesn't work
12:09atoHuh, it should.
12:09atoBut what you have now is fine.
12:09whimboomaybe that is because getInstallForFile resolves immediatelly to a vlaue
12:09atoMuch clearer.
12:10whimbooand as such the await doesn't wait for the resolve/reject lines insdie of then
12:10whimboobut just an idea... it's mystic
12:10whimboocool
12:11whimbooato, AutomatedTester: how should we prioritize bug 1395886?
12:11bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1395886 Marionette, normal, hskupin, ASSIGNED , Add support for proxy authentication (username / password)
12:11whimboogiven that no-one asked for it yet, make it P2 or P3?
12:18whimbooato: so I'm still confused about the desiredCapabilities and requiredCapabilities keys in capabilities. I thought also from last weeks meeting, that those do not exist anymore
12:19whimboowhy do we still handle it?
12:19atoWe support them because of backwards compatibility.
12:19atoAs far as Ive been able to tell from most the trace-logs in geckodriver issues, nearly all of our users still use desiredCapabilities.
12:20whimbooso its a relict from the old fx driver we have to keep maintaining?
12:21whimboomaybe it's time to stop it?
12:21atoWell not just FirefoxDriver, but all non-W3C driver implementations and Selenium too.
12:21whimbooi see
12:21atogeckodriver is AFAIK the only driver that supports firstMatch/alwaysMatch.
12:21atochromedriver may actually support it in a special mode now.
12:21whimbooI see
12:21whimbooso i will check a bit more but I think my bug is invalid then
12:22whimboowasn't clear when I checked the code
12:22atoI think in order to get rid of it we need to convince Selenium clients to send {capabilities: <firstMatch/alwaysMatch>}, possibly in addition to {desiredCapabilities: <caps>}.
12:23whimbooyeah
12:23atoThen that needs to stir for a while in releases before we can think about removing support for the legacy matching.
12:23whimbooso lets have a look here: curl -d &#39;{&quot;pageLoadStrategy&quot;: &quot;eager&quot;}&#39; http://localhost:4444/session
12:23whimboois that valid?
12:23atoThats not valid.
12:23whimbooi get a session created
12:23whimboowith pageload strategy not taken into acocunt
12:23whimbooand kept as normal
12:23atoYes, because we explicitly ignore top-level entries because we have to support desiredCapabilities/requiredCapabilities.
12:24atogeckodriver only checks the validity of the &quot;capabilities&quot; field at the top-level.
12:24whimbooshould such a case cause a session not created error?
12:24whimbooor do we simply ignore all those entries
12:24atoIndeed it should not, because Selenium clients send all kinds of shit in the JSON body.
12:25whimbooi wasn&#39;t aware of that.
12:25whimbooit makes it clear now
12:25whimboothanks. I will update the bug
12:25atoSo {&quot;capabilities&quot;: {&quot;alwaysMatch&quot;: {&quot;pageLoadStrategy&quot;: &quot;eager&quot;}}}, {&quot;capabilities&quot;: {&quot;firstMatch&quot;: [{&quot;pageLoadStrategy&quot;: &quot;eager&quot;}]}}, {&quot;desiredCapabilities&quot;: {&quot;pageLoadStrategy&quot;: &quot;eager&quot;}}, and {&quot;requiredCapabilities&quot;: {&quot;pageLoadStrategy&quot;: &quot;eager&quot;}} should be valid if I remember correctly.
12:26atoThis is a case of Postels Law.
12:27atoWe are being generous in accepting input for the old, crappy, non-specified input formats, but strict about what we accept for the new and standardised capabilities and in what we return.
12:30jgrahamI think rather than calling it Postel&#39;s law, it makes sense to say &quot;as required by the specification&quot;
12:30jgrahamIt&#39;s an intentional decision to allow arbitary top-level keys. It&#39;s not that we are being more liberal than required.
12:38atojgraham: I guess https://github.com/rust-lang/rfcs/issues/1621 would be nice to have in our case.
12:41atotype positive_i32 = i32 in 0..; // lower bound, but no upper bound
12:41atolet width = result.get(&quot;width&quot;).as_f64() as positive_i32;
12:41atoOr something.
12:46atos/f64/u64/ of course
12:46atos/u64/i64/
12:46atoMore tea required.
12:48wlachI don&#39;t suppose I could introduce anyone in fixing bug 1397213. should be just some basic mozprofile/mozregression hacking
12:48bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1397213 mozregression, normal, nobody, NEW , Can&#39;t install addons automatically
12:52whimbooato, jgraham: has one of you a second to explain what&#39;s wrong here in this Rust code?
12:52whimboohttps://irccloud.mozilla.com/pastebin/meooKIrv/
12:52whimbooi can&#39;t manage to fix it
12:54whimboooh wait... &Some() ?
12:55whimbooactually it looks like I have to use `let &Some(ref runner) = browser`
13:01whimboojgraham: try_wait!() seems to work!
13:01whimboofor checking the process status
13:01jgrahamwhimboo: Great
13:02atowhimboo: I think at that point you always have a FirefoxRunner, right? So you dont need to pass in an Option<FirefoxRunner> but could error much further up if a FirefoxRunner hasnt been created.
13:03atoBut maybe Im mis-remembering how this works.
13:04armenzgahal: was it bitwarden the password manager you kept talking about?
13:06whimbooato: the browser is created in a different method of the same class. So i feel better in passing in an option here, because that already is what we store on the MarionetteConnection class as self.browser
13:07atoIf its the same uh, class you probably dont need to pass it at all.
13:12ahalarmenzg: yeah
13:15armenzgahal: OK I will switch over; LastPass does not work for Nightly
13:15armenzgI hope I can import
13:15ahalyou can, quite easily
13:15ahalthere&#39;s a wizard iirc
13:16armenzg_brbty
14:38davehuntekyle: ActiveData appears to be unavailable
14:38ekyledavehunt:
14:38ekyledavehunt: will look..
15:06ekyledavehunt: sorry for the delay... fixed
15:07ekyledavehunt: the fix was a simple bounce of a non-data node, which should have been the first thing I tried :/
15:08ekyledavehunt: the newer versions of ES are claimed to be more stable.
15:09davehuntekyle: awesome, thanks!
15:09davehuntekyle: is there any monitoring for the service?
15:19ekyledavehunt: I get email alerts; the problem is I was in transit this morning, and then in a meeting.
15:20ekyledavehunt: the next step is to automate procedure to fix this type of failure
15:20davehuntekyle: okay, cool
15:22whimbooato: not the same class, sorry.
15:49emorleyemceeaich dylan|afk: my bugzilla footer has just started showing the saved search &quot;JS JIT Bugs in 57 Cycle&quot; even though I didn&#39;t select it. Any ideas why? :-)
15:50RyanVMsame here
16:15atowhimboo: I will continue work on the element/window rect typing tomorrow, I think.
16:17atoWill squeeze it in.
16:34globemorley, RyanVM: urgh, looks like someone with rights doesn&#39;t know how to use them. can you file a bug?
16:37* glob files
16:40globbug 1397353
16:40bugbotBug https://bugzilla.mozilla.org/show_bug.cgi?id=1397353 Administration, normal, nobody, NEW , &quot;JS JIT Bugs in 57 Cycle&quot; search added to everyone&#39;s footer
16:51emorleyty
16:54emorleycamd AutomatedTester https://bugzilla.mozilla.org/show_bug.cgi?id=1377528
16:54bugbotBug 1377528: Treeherder, normal, nobody, NEW , Publish notification about completed try push to pulse
16:56AutomatedTestermaja_zf: is ^ still valid
16:56AutomatedTesterdoes the info from from TC give you what you need?
16:58maja_zfAutomatedTester: invalid, I think TC alone is enough now
17:09emceeaichemorley: does that saved search still show up in your footer?
17:15emorleyemceeaich: I manually unticked it earlier
17:50RyanVMemceeaich: fixed here
18:49whimbooRyanVM: hi. are you aware of a bug on OS X which doesn&#39;t stop audio from a playing video when navigating back?
18:50RyanVMnot offhand, sorry
18:50whimbooi see it in both Nightly and Beta
18:50whimbook, will check then
19:20gpsahal: thank you for converting `mach python-test` to pytest. so much nicer.
19:21ahalNp!
19:21butter... sitting in a tree... k-i-s-s-i-n-g!!
19:22butter:D <3<3<3<3<3<3<3 love is in the air!!
23:54emceeaichjmaher|afk: hopefully I did not muddy the waters with my answer to andym
7 Sep 2017
No messages
   
Last message: 15 days and 10 hours ago