mozilla :: #developers

13 Sep 2017
00:00Asaand all the less bad 5 and 10 second pauses I would get with less scrollback feel pretty instant.
00:09pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/08a75460a0d8 - Wes Kocher - Backed out 4 changesets (bug 1352016) for gfx crashes in test_TelemetryEnvironment.js a=backout CLOSED TREE
00:20RyanVM|bblAsa: sweeeet
00:22AsaRyanVM|bbl: yeah. I think this bug was kind of linear with how big your scrollback is and I've got a strange usage pattern with huge scrollback but the fix will make it feel faster for regular folks too I think.
00:22RyanVM|bbllove how simple the patch is too
00:22AsaI wish we could find 10x perf improvements in top 10 websites every day
00:23RyanVM|bbl+1
00:23AsaGecko Profiler FTW on this one.
00:41* John-Galt looks at treestatus, whistles idly...
00:42pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/f4b0f3105c72 - Wes Kocher - Bug 1398869 - Now re-fix the non-stylo tests a=bustage CLOSED TREE
00:48KWiersoJohn-Galt: yeah
00:49KWiersoinfra broke a bunch of things today. While infra was broken, all windows builds were also simultaneously broken for a dozen or so pushes
00:49KWiersoand, shockingly, while all windows builds were simultaneously broken for those dozens or so pushes, a bunch of actual test failures crept in
00:49KWiersoon windows
00:49KWiersowith no clear cause
00:52John-GaltWindows usually doesn't need a reason.
00:54KWiersoJohn-Galt: happily, only some of the several failures have jobs with > 40 minute runtimes
00:55KWiersoif it makes you feel any better, I think I'll have inbound opened soonish
00:55John-Galt\o/
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/cdfda62c6eff - jason laster - Bug 1398893 - Breakpoints cannot be removed. r=jdescottes, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/aa3740564226 - Zibi Braniecki - Bug 1397925 - Refresh OSPreferences mSystemLocales when intl.locale.os changes. r=rnewman, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/bb89faa93b22 - Shane Caraveo - Bug 1385630 - Fix extension sidebar restore after app update. r=Gijs, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/bd8b51b80738 - Daniel Stenberg - Bug 1359624 - Disable nsAuthSambaNTLM module on OSX. r=mayhemer, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/b55ee75e7aba - Steve Fink - Bug 1389974 - "Fix" false positive rooting hazard in ArrayInfo(). r=jonco, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/e35824dc0782 - Hiroyuki Ikezoe - Bug 1334582 - Use UINT64_MAX instead of IsInfinite() for checking whether TimingParans.mIterations is infinite or not. r=boris, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/a18761befb03 - Hiroyuki Ikezoe - Bug 1334582 - Check whether overall progress exceeds UINT64_MAX. r=boris, a=lizzard
00:59pulsebotCheck-in: https://hg.mozilla.org/releases/mozilla-beta/rev/867b9e3b4f92 - Richard Newman - Bug 1378501 - Tell Gecko when the OS locale changes, even when backgrounded. r=mcomella, a=lizzard
01:06pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/8db97b50b5d3 - Wes Kocher - Backed out changeset 5ed9b657a1ce (bug 1352459) for xperf failures a=backout CLOSED TREE
01:10KWiersoJohn-Galt: there you go
01:10John-Galt\o/
01:10John-GaltThanks :)
01:11pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/237a3f842dc1 - Andrew Halberstadt - Bug 1398765 - Fix directory exclusion bug in py-compat linters, r=gps
01:12pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/776a65d43a5e - Ted Campbell - Bug 1398601 - Support target objects in js::ExecuteInJSMEnvironment r=jorendorff
01:12pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/eddc38b0afd0 - Ted Campbell - Bug 1398601 - Add global/non-syntactic prefix to subscript loader cache r=kmag
01:12pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/f5c26c3407c0 - Ted Campbell - Bug 1398601 - Fix subscript loader when using JSM global sharing r=kmag
01:12pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/05957a92b1a5 - Ted Campbell - Bug 1398601 - Add testcase for loading subscripts within a JSM r=kmag
01:12pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/b728872f4d9a - Ted Campbell - Bug 1398601 - Don't allow NSVO in js::CreateObjectsForEnvironmentChain r=jorendorff
01:13pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/9a676cb6f09a - Kris Maglione - Bug 1399005: Always dispatch OnStop from the IO thread. r=mixedpuppy
01:14gandalfkhuey: ping
01:14pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/82f171552905 - Ed Lee - Bug 1399226 - Add high-res icons, image placeholders and bug fixes to Activity Stream. r=dmose
01:15pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/0aff84a0a8af - Michael Comella - Bug 1388396: Increase max favicon scale factor to 3. r=sebastian
01:15pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/8756322ebb4a - Michael Comella - Bug 1388396: Add IconRequestBuilder.forActivityStream. r=sebastian
01:15pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/77f982a2f17b - Michael Comella - Bug 1388396: Use IconRequestBuilder.forActivityStream in AS UI. r=sebastian
01:15pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/44fa7b379742 - Michael Comella - Bug 1388396: Add MinimumSizeProcessor; .forActivityStream takes effect. r=sebastian
01:15pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/3300c15011d3 - Michael Comella - Bug 1388396 - review: Add TestMinimumSizeProcessor. r=mcomella
01:17pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/cdaed5454569 - Andr Bargull - Bug 1398226 - Remove the indirection through NewDependentString when creating inline-strings in StaticStrings::getUnitStringForElement. r=jandem
01:17pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/781bc2a066f2 - Jan Keromnes - Bug 1398082 - Support compiling stylo with llvm-config-5.0 or llvm-config-6.0. r=froydnj
01:17pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/5b700335fc30 - Mike de Boer - Bug 1374749 - Animate the panelviews differently to make it look as if the view to show is pushing the previous view out of the panel. r=jaws
01:18pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/b8cb090f952b - Thom Chiovoloni - Bug 1399191 - Use an arrow function to avoid an undefined this value when calling namedTimer. r=markh
01:19pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/2a131a0c2d11 - Julian Descottes - Bug 1399180 - select devtools tab on mousedown rather than on click;r=gl
01:20pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/4ab54f3ebbc5 - Mark Striemer - Bug 1390647 - Match about:debugging and about:addons sidebar to about:preferences r=jdescottes
01:20pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/c8cb0cd2d1e5 - Bobby Holley - servo: Merge #18465 - Eliminate RestyleData (from bholley:kill_restyledata); r=emilio
01:20pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/7d9f5239adc3 - Anthony Ramine - servo: Merge #18459 - Kill ComputedValueAsSpecified (from servo:kill-cvas); r=emilio
01:20pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/4665231558c0 - Julian Seward - servo: Merge #18460 - Bug 1398593 - stylo: mitigate performance impact of fallible allocati (from julian-seward1:master); r=emilio
01:22pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/fc635ab83267 - Emilio Cobos lvarez - Bug 1397983: Rename nsXBLBinding::{Install,Uninstall}AnonymousContent to {Bind,Unbind}AnonymousContent. r=smaug
01:22njnso, is it "iPhone ex" or "iPhone ten" ?
01:22pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/d928c61a691f - Emilio Cobos lvarez - Bug 1397983: Synchronously bind/unbind XBL anonymous content from the bound content's Bind/UnbindFromTree. r=smaug
01:23pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/a98262b2acc1 - Bas Schouten - Bug 1399247: Limit the scope of D3D11MTAutoEnter in order to avoid holding it while grabbing the D2D lock. r=mchang
01:25pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/be5cbdf6f420 - Andi-Bogdan Postelnicu - Bug 1379961 - Add platform and architecture name to MozbuildObject. r=gps
01:26pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/9fbfe01ae6f1 - Chenxia Liu - Bug 1380811 - Add telemetry to collect boolean state for toggle preference. r=francois,mcomella
01:27pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/afaf910a6aca - Dan Glastonbury - Bug 1399009 - Update cubeb-pulse-rs to a386d91... r=kinetik
01:28pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/247dc838b340 - Ryan VanderMeulen - Bug 1367895 - Remove some obsolete nightly_only & release_or_beta test manifest annotations. r=mconley
01:28pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/b21cba0005ef - Benjamin Bouvier - Bug 1370886 - Test for source script offset behavior in asm.js. r=luke
01:28pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/9367b7801b70 - Benjamin Bouvier - Bug 1399239 - Remove unused srcBodyStart. r=luke
01:28pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/1cfe6ff24019 - Michael Brennan - Bug 1347827 - Add Persist Logs checkbox in NetMonitor toolbar. r=Honza
01:28Asanjn: pretty sure it's ten
01:28pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/d5f4464e0054 - Thom Chiovoloni - Bug 1394158 - Make BrowserUtils.getSelectionDetails check for URLs inside editable elements r=Felipe
01:29njnAsa: ok, thanks
01:29* njn is curious to see them making that mistake again
01:29pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/fdbf2ccf813e - Andr Bargull - Bug 1398751 - Part 1: Add fast-path for typed arrays in js::GetElements to speed-up Function.apply with typed arrays. r=evilpie
01:29pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/f0f98cfb1a45 - Andr Bargull - Bug 1398751 - Part 2: Add fast path to convert int32 value to a code point. r=evilpie
01:30pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/1ca25d95e1aa - Andrew McCreight - Bug 1216885 - Make nsISelectionPrivate not inherit from nsISelection. r=smaug
01:31pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/ce559fa15078 - ahillier - Bug 1399200 - Don't stop loading thumbnail when image redirects r=Mardak
01:32pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/f05ca5366731 - Grigory Kruglov - Bug 1368147 - Pre: tighten up access levels in AndroidFxAccount r=sebastian
01:32pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/487a253f5548 - Grigory Kruglov - Bug 1368147 - Pre: remove unused code r=sebastian
01:32pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/0ca257bd9dcb - Grigory Kruglov - Bug 1368147 - Pre: Don't forget to call super.finalize in finalize override r=sebastian
01:32pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/d881fed7d5ca - Grigory Kruglov - Bug 1368147 - Migrate sync and reading list preferences to be keyed by account UID r=nalexander,sebastian
01:32pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/3044b28ca2ba - Grigory Kruglov - Bug 1368147 - Support renaming an account in response to profileUpdate events r=nalexander,sebastian
01:33pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/a230d92e4e94 - Matthew Noorenberghe - Bug 306730 - Do not include the token name in prompts for the internal key slot. r=keeler
01:34pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/2065eef5fa19 - Jing-wei Wu - Bug 1380266 - Keep identical click behavior between HomeAsUpIndicator and back button. r=nechen
01:35pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/588f15fdda38 - Jing-wei Wu - Bug 1398657 - Part 1: Remove mdpi drawables. r=nechen
01:35pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/9dcc9c6cb581 - Jing-wei Wu - Bug 1398657 - Part 2: Remove redundant drawables. r=nechen
01:36pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/eee73f48e071 - Brian Grinstead - Bug 1398760 - Flex align the console filter checkbox to correct vertical misalignment;r=Honza
01:36pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/0e8adec38eb8 - Brian Grinstead - Bug 1398760 - Prevent text selection in console filter checkbox;r=Honza
01:37pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/d72893703bf7 - Michael Comella - Bug 1398834: Add pin/unpin context menu strings. r=liuche
01:37pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/885dbb7d0b17 - Michael Comella - Bug 1398834: Add Pin to Top Sites page menu item. r=liuche
01:37pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/cabd92c1a177 - Michael Comella - Bug 1398834: Add telemetry for new browser menu pin site. r=liuche
01:38pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/a104181ac18b - James Cheng - Bug 1393710 - Add thread name into NS_DebugBreak for better debugging. r=froydnj
01:40pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/f2bd9e8d0975 - Michael Kelly - Sync Bug 1399245: Fix mozilla/normandy#1036: Remove OSX-specific close button CSS for Heartbeat bar. r=Gijs
01:40pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/9771577eeeae - Michael Kelly - Sync Bug 1399245: Fix mozilla/normandy#1058: Bind PreferenceExperiment driver functions. r=Gijs
01:42pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/c9589184e49f - James Cheng - Bug 1398102 - [Part1] canPlayType should return 'maybe' if the codec parameter is empty. r=gerald
01:42pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/3dbe64e08fd0 - James Cheng - Bug 1398102 - [Part2] Fix test fail due to wrong assumption. r=gerald
01:47njnoh, have the perma-red Windows builds been removed from try?
01:50pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/fa2ccc95cf0c - Ethan Lin - Bug 1393077 - Part1. Round the transformed rectangles and transformed points in StackingContextHelper. r=kats
01:50pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/fc237de487fd - Ethan Lin - Bug 1393077 - Part2. Round the offset for the fallback. r=kats
01:50pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/fa8562b32dae - Ethan Lin - Bug 1393077 - Part3. Modify annotations for affected testcases. r=kats
02:00dmajornjn: maybe https://bugzilla.mozilla.org/show_bug.cgi?id=1384706 ?
02:00njndmajor: looks plausible, thanks
02:08pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/11a7e286a5b7 - Oriol Brufau - Bug 1395313 - Let JSON Viewer ignore the BOM. r=Honza
02:25pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/0337d3437615 - Jeff Muizelaar - Bug 1392523. Remove duplicated gfxContext* parameter. r=mattwoodrow
02:30pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/819875ef9f03 - Chris Pearce - Bug 1399309 - Make nsTSubstring.cpp include MathAlgorithms.h. r=froydnj
02:31pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/b393da8b7c59 - Chris Pearce - Bug 1399308 - Include nsThreadUtils.h in nsISupportsImpl.cpp. r=froydnj
02:31pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/9e6a4d96cf76 - Boris Zbarsky - Bug 864845 part 1. Stop using 'content' in devtools tests. r=ochameau
02:31pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/c991b71d3642 - Boris Zbarsky - Bug 864845 part 2. Make window.content chromeonly. r=mystor
02:32pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/080b044b13e3 - Mark Hammond - Bug 1395332 - perform some post-profile-migration steps to ensure Sync works correctly in the new profile. r=Gijs
02:34pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/8e8f901c19ca - Glenn Watson - servo: Merge #18470 - Update WR (backface API changes) (from glennw:update-wr-bf); r=pcwalton
02:40gandalfallstarschh: ping
02:41allstarschhgandalf: pong
02:41gandalfallstarschh: would you have a moment to unblock me with bug 1397365? I just spent the whole day in firstPartyDomain world and need a bit of a push :)
02:43allstarschhgandalf: do you know why the flag LOAD_FLAGS_DISALLOW_INHERIT_PRINCIPAL is gone?
02:46pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/de03f659d5e6 - Jeff Muizelaar - Revert "Bug 1392523. Remove duplicated gfxContext* parameter. r=mattwoodrow"
02:46pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/5291605468b8 - Jeff Muizelaar - Bug 1398959. Remove duplicated gfxContext* parameter. r=mattwoodrow
02:46gandalfI have a faint idea
02:46gandalfallstarschh: basically, I'm working on test http://searchfox.org/mozilla-central/source/browser/components/originattributes/test/browser/browser_firstPartyIsolation_js_uri.js#15
02:47gandalfwithout my patch, if I understand it correctly, we open a new window, which by default gets opened with "about:blank". This default "about:blank" gets the firstPartyDomain set to some UUID.mozilla
02:47gandalfthen we loadURI "javascript:1;" and the firstPartyDomain ID stays
02:48gandalfwith my patch, we *do not* load any URL by default.
02:48gandalfor at least, we shouldn't.
02:48gandalfso when we loadURI("javascript:1;"); the firstPartyDomain is empty
02:48gandalfand the test fails
02:48pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/d7af29b83ac3 - Julian Descottes - Bug 1399028 - photon update - devtools icon, tab and accordion colors;r=gl
02:48pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/c5aedf62e0af - Julian Descottes - Bug 1399028 - fix assert messages in browser_theme.js test;r=gl
02:50gandalfallstarschh: my understanding is that in the test we want now to generate the firstPartyDomain UUID when we do the loadURI, but for that we need to pass some flags.
02:51gandalfI tried to replace the loadURI with:
02:51gandalf browser.loadURIWithFlags(`javascript:1;`,
02:51gandalf Ci.nsIWebNavigation.LOAD_FLAGS_DISALLOW_INHERIT_PRINCIPAL |
02:51gandalf Ci.nsIWebNavigation.INTERNAL_LOAD_FLAGS_FIRST_LOAD, null, null, null)
02:51gandalfbut either this is not enough or javascript:1; URI just doesn't generate the firstPartyDomain UUID :(
02:52gandalfdoes my guesswork sound anyhow reasonable?
02:52allstarschhgandalf: the problem is about:blank, as I pointed out in https://bugzilla.mozilla.org/show_bug.cgi?id=1397365#c42
02:53allstarschhgandalf: javascript: URI just inherits from about:blank, so if about:blank is wrong, javascript: URI will also wrong
02:53gandalfallstarschh: yes, so without my patch the &quot;about:blank&quot; is not really loaded, it&#39;s just intrinsically there because by default <browser> in XUL creates it
02:53gandalfwe want to remove the intrinsic about:blank
02:53gandalfso the <browser nodefaultsrc=&quot;true&quot;> does not start with about:blank at all
02:54gandalfso the first URL that is loaded should set the firstPartyDomain.
02:54gandalfdoes it make sense?
02:55gandalfI see that in the test, it expects the firstPartyDomain to be sth like &quot;UUID.mozilla&quot;, and if I load a regular &quot;about:blank&quot; or &quot;about:mozilla&quot; its set to &quot;about.UUID.mozilla&quot;. Can you tell me what flags do I need to pass to docShell&#39;s loadURI to make it generate &quot;UUID.mozilla&quot; type firstPartyDomain ?
02:57gandalfor, maybe it&#39;s ok if firstPartyDomain is empty in this test?
02:58gandalfsince we don&#39;t have the about:blank before it, javascript: URL should keep the firstPartyDomain empty?
02:58allstarschhgandalf: we shouldn&#39;t pass any flag to it, the problem is why the flag &#39;LOAD_FLAGS_DISALLOW_INHERIT_PRINCIPAL&#39; is gone with your patch
02:59gandalfbecause instead of <browser> (which inherintly loads about:blank) we now have <browser nodefaultsrc=&quot;true&quot;> which doesn&#39;t inherently load anything.
02:59gandalfthen in BrowserTestUtils.openNewBrowserWindow we did insert the &quot;about:blank&quot; URL to load, but that&#39;s different. I updated the patch today
03:00gkwis about:newtab a privileged URL wrt. WebExtensions? I seem to be unable to activate it, but other non-privileged URLs are fine, e.g. about:blank
03:00gandalfso now, you can do BrowserTestUtils.openNewBrowserWindow({ remote: true }, &quot;&quot;); and the second argument is the default URL to load. Which can stay empty
03:00gkwhttps://developer.mozilla.org/en-US/Add-ons/WebExtensions/API/tabs/update
03:01gandalfallstarschh: my guess is that when the &quot;about:blank&quot; come from <browser> it has LOAD_FLAGS_DISSALOW_INHERIT_PRINCIPAL. And when it comes from arguments to WindowWatcher.openWindow, it doesn&#39;t.
03:02gandalfbut now I think that maybe it&#39;s ok. If we open a window without any URL loaded into the first tab, and the first URL we load is &quot;javascript:1&quot; maybe firstPartyDomain should be empty?
03:04allstarschhgandalf: firstPartyDomain shouldn&#39;t be empty
03:06gandalfhmm
03:06gandalfok
03:06gandalfso, how does it work for all other tabs?
03:07gandalffor all but the first tab in each window
03:07gandalfwe create a new <browser> element with nodefaultsrc=&quot;true&quot; - http://searchfox.org/mozilla-central/source/browser/base/content/tabbrowser.xml#2261
03:07gandalfand that&#39;s exactly what I&#39;m trying to do for the first <browser> in the window as well
03:08gandalfhow is the firstPartyDomain calculated for all those other tabs?
03:08gandalfsince they don&#39;t get the about:blank
03:09allstarschhthe flag should come from http://searchfox.org/mozilla-central/source/dom/ipc/TabChild.cpp#1158,
03:09gandalfallstarschh: I see this in tabbrowser.xml&#39;s addTab - http://searchfox.org/mozilla-central/source/browser/base/content/tabbrowser.xml#2796
03:10gandalfso, when we open a new tab, we create a new browser with nodefaultsrc=&quot;true&quot;, and loadURIWithFlags into it.
03:11gandalfwhen we create a window, the first browser doesn&#39;t have the nodefaultsrc and I&#39;m trying to add it. How can I replicate the same behavior for the first tab as it has for other tabs?
03:11tcampbellis there a sheriff around? I need to be backed out from autoland
03:12pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/a45742d015d3 - Phil Ringnalda - Backed out 5 changesets (bug 1398601) for unexpected crashtest assertions
03:12philortcampbell: sorry, I&#39;m busy backing out tcampbell
03:12tcampbelloops
03:12tcampbellphilor: thanks
03:13philornp, thanks for spotting it too
03:14tcampbellthat was attempt #2 at this patch set =\
03:14tcampbellso many edges in the script loader
03:16tcampbelloh jeez.. I see what went wrong. I guess that isn&#39;t a hard error in any other config
03:18allstarschhgandalf: I need to check what nodefaultsrc does in dom, will reply on bugzilla later
03:18gandalfallstarschh: ok, one more bit of info
03:18pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/649a61cd9381 - Fischer.json - Bug 1398987 - Can&#39;t see the onboarding overlay close button with Windows dark high contrast modes, r=rexboy
03:18gandalftoday&#39;s patch changes a couple things. the BrowserTestUtils.openNewBrowserWindow now has a second parameter which is passed to WindowWatcher.OpenWindow as the URL to load
03:19gandalfthose are the results:
03:19gandalfhttps://pastebin.mozilla.org/9032275
03:19philortcampbell: apparently mochitest-chrome too, https://treeherder.mozilla.org/logviewer.html#?job_id=130504190&repo=autoland - I thought only reftest harness did assertion counting, but apparently not
03:20allstarschhgandalf: okay, I&#39;ll try your latest patch
03:20gandalfallstarschh: my guess is that there&#39;s something special happening when the about:blank comes from <browser> rather than loadURI that constructs the ID differently
03:20gandalfok, thank you! :)
03:20allstarschhnp
03:21tcampbellphilor: ooh, good to know. I&#39;ll watch for those in try
03:30Mardakanyone know where to get https://people.mozilla.org/~jmaher/taloszips/zips/tp5n.zip now?
03:34philor&quot;Got Namast, expected Namast&quot;
03:34philordevtools *expects* mojibake?
03:35bcMardak: you can get it from tooltool.
03:35bcIf you do an ./mach awsy-test --quick it will download it for you.
03:36Mardakaha! thanks
03:37John-GaltFun fact: Searching for Namast breaks Searchfox
03:39* philor pats dxr on its mangy head
03:44philormmm, and that failure may be relevant to tcampbell&#39;s interests
03:45bcMardak: it will be in _tests/awsy/html/page_load_test under your objdir
03:45Mardakbc: yup. got it! thanks again
03:46Mardakanyone know how to get tp5n files changed? I see that some urls are changed to be httpdisabled:// to prevent network connections, but not all
03:46bcMardak: file a bug under Testing:Talos would be my first step. jmaher will help.
03:47Mardakok i ni? him already but i suppose it&#39;ll need to be fixed in a separate bug anyway
03:47Mardakthanks again :)
03:48tcampbellhmm
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/4ec271cc83a0 - Kris Maglione - Bug 1398974: Part 1 - Add webRequest listeners to webext talos add-on. r=mixedpuppy,jmaher
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/959398912fee - Kris Maglione - Bug 1398974: Part 2 - Add tab listeners to webext talos add-on. r=mixedpuppy,jmaher
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/28b7ba19f12b - Kris Maglione - Bug 1398974: Part 3 - Add pageAction to webext talos add-on. r=mixedpuppy,jmaher
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/e3cb7f8e1174 - Kris Maglione - Bug 1398974: Part 4 - Add browserAction to webext talos add-on. r=mixedpuppy,jmaher
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/52e225228c3e - Kris Maglione - Bug 1398974: Part 5 - Add content script to webext talos add-on. r=mixedpuppy,jmaher
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/59069fc4ad51 - Kris Maglione - Bug 1398974: Part 6 - Add content script messaging to webext talos add-on. r=mixedpuppy,jmaher
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/6ff73d0599c5 - Kris Maglione - Bug 1398974: Part 7 - Update and sign talos webext extension. r=me
03:52pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/4f6d9ea5a49e - Kris Maglione - Bug 1398974: Follow-up: Disconnect StreamFilters when closing extension context. r=me
03:58tcampbellphilor: that Namaste issue is also me
03:58tcampbellI can repro locally off central
03:58tcampbell(with my patches on it)
03:59philornamaste
03:59* philor stops trying to persuade the worthless infra to backfill the gap
03:59pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/a8d6f6fd7bda - Andrea Marchesini - Bug 1359017 - onmessageerror for MessagePort in case StructuredClone algorithm fails when deserializing, r=masayuki, r=smaug
03:59pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/354139a05e75 - Andrea Marchesini - Bug 1359017 - onmessageerror for BroadcastChannel in case StructuredClone algorithm fails when deserializing, r=masayuki, r=smaug
03:59pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/9248962260f9 - Andrea Marchesini - Bug 1359017 - onmessageerror for Workers in case StructuredClone algorithm fails when deserializing, r=masayuki, r=smaug
03:59pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/7b45c1aebf7f - Andrea Marchesini - Bug 1359017 - onmessageerror for WindowEventHandlers in case StructuredClone algorithm fails when deserializing, r=masayuki, r=smaug
03:59pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/43d65c965f0d - Andrea Marchesini - Bug 1359017 - WPTs for onmessageerror, r=masayuki
04:00gandalfJohn-Galt: I&#39;m wondering if it&#39;s possible that the XULFrameLoaderCreated is fired too early if <browser nodefaultsrc=&quot;true&quot;> is used. Or the fact that we don&#39;t load any document means that target.messageManager is going to stay null until we load some URL into the tab?
04:01John-Galttcampbell: With the cache path issue fixed or not?
04:01tcampbellJohn-Galt: with it fixed
04:01John-Galtgandalf: Seems likely
04:01John-Galttcampbell: Interesting. But not without your patches?
04:01* tcampbell tests that
04:02John-GaltThe devtools do use the subscript loader to load their code, and it takes an optional charset. If you don&#39;t pass one, you get something like 8859-1. It&#39;s entirely possible that they&#39;re loading the same scripts in multiple ways, though, and it currently only works because the correct one is cached when they try the incorrect one.
04:03* John-Galt has seen that happen
04:03John-GaltThis, for instance, is right: http://searchfox.org/mozilla-central/source/devtools/shared/base-loader.js#232
04:03John-GaltWhile this is wrong: http://searchfox.org/mozilla-central/source/devtools/server/main.js#78
04:04John-GaltAnd the second one probably loads into a global object, while the first loads into a plain object.
04:04tcampbellfun..
04:05John-GaltHm. Also here: http://searchfox.org/mozilla-central/source/devtools/shared/worker/loader.js#273
04:06John-GaltOh, no, not that one.
04:07John-Galttcampbell: Anyway, if you can produce it locally only with your patches, maybe log when it&#39;s loaded and what charset, and see if it varies.
04:07philorah, so I had it right, they do (or rather, should) expect mojibake
04:08* tcampbell wishes icecream didn&#39;t flake out so much
04:09John-GaltI don&#39;t think so. I think they expect non-mojibake, but one of their scripts is being loaded with the wrong charset.
04:09* John-Galt has had very little trouble with icecream lately.
04:09John-GaltExcept when I try to use it at an office. There&#39;s always one machine that doesn&#39;t work, but keeps getting jobs sent to it.
04:11tcampbellmy jobs gets dispatched to all the machines, but then only one seems to continue to process, but it is slower than mine =S
04:11tcampbellI think it got worse when I upgrade GCC version
04:12tcampbellI&#39;m not sure if all the new warnings that are on by default affect things
04:12John-GaltOh, I switched to clang ages ago
04:13John-Galttcampbell: Hm. I get that failure with a local inbound build, without your patches.
04:14tcampbellodd.. I had it with central from a few days ago + the patches of that bug
04:17tcampbellJohn-Galt: philor: m-c from sunday fails that test for me
04:18tcampbelli&#39;m not sure where source of problem is
04:20John-GaltThose two scripts are definitely loaded with different charsets... http://searchfox.org/mozilla-central/source/testing/mochitest/browser-test.js#771-772 vs. http://searchfox.org/mozilla-central/source/devtools/shared/base-loader.js#232
04:23philordepends on running a previous test in the chunk to misset a pref and leave it set?
04:23John-GaltAnd adding a charset to the one in browser-test.js fixes it.
04:24John-Galtphilor: Hm... possibly.
04:33gandalfJohn-Galt: where can I find the code for `browser.windows.create()` used in the test extension?
04:34John-Galtgandalf: http://searchfox.org/mozilla-central/source/browser/components/extensions/ext-windows.js
04:34gandalfty!
04:36John-Galttcampbell: Ah. It fails if I run it locally because it&#39;s normally skipped, but if you specify the whole path it runs.
04:37tcampbellohh.. so why is it not skipped anymore?
04:37John-GaltIt&#39;s skipped in e10s mode, anyway
04:38John-GaltIt passes for me locally with e10s disabled
04:38philoroh, that would explain why it was only on win7
04:41John-GaltHeh. And if I add the correct encoding to browser-test.js with e10s enabled, it fails with: 92 INFO TEST-UNEXPECTED-FAIL | devtools/client/commandline/test/browser_gcli_async.js | member[1]: predictions (for &#39;tsslo<TAB>&#39;) - Got Namast, expected Namast
04:41John-GaltSo apparently in e10s mode it&#39;s somehow relying on the encoding getting mangled the same way as it does in the test file.
04:42tcampbellThe startup cache is distinct for the child and parent process, right?
04:43pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/ae71d517c443 - Nils Ohlmeier [:drno] - Bug 1396974: stop bundeling inactive m-sections. r=bwc
04:43pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/06c3517b0f33 - Nils Ohlmeier [:drno] - Bug 1396974: make tests ensure no bundle-only is present in inactive m-swctions. r=bwc
04:43John-Galttcampbell: Complicated. The regular startup cache isn&#39;t used in the child. The script preloader startup cache tries to use the same cache for both the parent and the child.
04:43John-GaltIf the same scripts are used in both.
04:44tcampbellJohn-Galt: I noticed there are two files in profile though
04:44John-GaltYeah, but if a script exists in the child cache, the parent loader uses that instead of its own.
04:46pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/834aca219fbb - Scott Wu - Bug 1386922 - Add learn more link to autofill address, r=lchang,MattN
04:46pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/70d1f53f21f2 - Scott Wu - Bug 1386922 - Add learn more link to autofill credit card, r=lchang,MattN
04:47tcampbellJohn-Galt: sounds like when I broke subscript cache this problem appeared similar to it&#39;s e10s problem. Seems they really need to fix the test case
04:47pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/84abfad4ada9 - JW Wang - Bug 1398711. P1 - remove unused methods. r=gerald
04:47pulsebotCheck-in: https://hg.mozilla.org/integration/autoland/rev/b420bb292e68 - JW Wang - Bug 1398711. P2 - write to mClosed only when the cache monitor is held. r=gerald
04:47John-Galttcampbell: Ah, suspicion confirmed. The head.js for that test loads mockCommands.js into its global using the subscript loader, but with the wrong charset. The devtools base-loader.js later loads it with the right charset, into a non-syntactic object scope.
04:47John-GaltAnd with your patches, they use different cache entries.
04:47tcampbellfun
04:48John-GaltSo I think we need to fix that head.js to use the right encoding, and same for browser-test.js
04:48John-GaltWhich hopefully won&#39;t break anything else that depends on broken charsets :/
04:48John-GaltThat... or just change those files to use \u escapes.
04:48John-GaltWhich would be simpler.
04:49John-GaltI really hate when fixing bugs winds up breaking things.
04:50tcampbellyeah.. we could roll back the cache change add add the same copy-script-to-non-syntactic stuff and revisit later, but that doesn&#39;t feel better \u escapes
04:51John-GaltRolling back the cache change so some test can continue to rely on the broken behavior seems like not a good solution.
04:51John-Galt\u escapes are probably the best thing, for now. Then fixing the test harness to load things in utf-8
04:52tcampbellI should be in bed. Will continue in the morning
04:52John-GaltYeah, same. &#39;night
04:54pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/a7877592dff7 - Kris Maglione - Backed out changeset 6ff73d0599c5 (bug 1398974) for still being flaky on Winodws. r=backout
04:58pulsebotCheck-in: https://hg.mozilla.org/integration/mozilla-inbound/rev/0657b94b6924 - Christoph Kerschbaumer - Bug 1397652 - Update tests within browser/ to comply with new toplevel data: URI navigation policy. r=gijs
05:02gandalfJohn-Galt: so, here: http://searchfox.org/mozilla-central/source/browser/components/extensions/ext-windows.js#186
05:02gandalfthe firect onXULFramLoaderCreated fired line above works on central and with my patch
05:02gandalfbut line 193 is setting a handler that is triggered by tabbrowser.xml&#39;s updateBrowsderRemoteness
05:03* John-Galt can believe that
05:03gandalfand with nodefaultsrc this one doesn&#39;t have messageManager
05:03John-GaltThat was my first suggestion, actually.
05:03John-GaltHm
05:04gandalfI&#39;m a bit lost as to why is that, but my understanding of browserremoteness is limited
05:04gandalfI know that if I pause the test
05:04gandalfand open browser console, the linkedBrowser for the tab does have the messageManager
05:05John-GaltMy guess is that with nodefaultsrc we just skip initializing the frameloader until we have a URL to load.
05:05John-GaltOr it would be, if I weren&#39;t already suspicious about the Quitter bugs.
05:05gandalfbut we do have a url to load. We&#39;re loading dummy.html
05:05John-GaltBecause those suggest we never initialize the message manager at all
05:05gandalf<