mozilla :: #developers

18 May 2017
00:08pulsebotCheck-in: - Drew Willcoxon - Bug 1303781 - Sometimes selecting a history entry opens a Google search instead of navigating to the entry. r=mak
00:33pulsebotCheck-in: - Masatoshi Kimura - Bug 817374 - Fix inconsistency of the setOverLink parameter. r=mossop
00:35pulsebotCheck-in: - Andrew Swan - Bug 1365420 Re-enable and fix test_ext_permissions.js r=zombie
00:36pulsebotCheck-in: - Mike Hommey - Bug 1365194 - Call moz_abort directly instead of using a macro to override abort. r=njn
00:36pulsebotCheck-in: - Mike Hommey - Bug 1365194 - Remove MOZ_JEMALLOC_API from _malloc_options and _malloc_message. r=njn
00:36pulsebotCheck-in: - Mike Hommey - Bug 1365194 - Make `extern "C"` part of MOZ_MEMORY_API and MOZ_JEMALLOC_API. r=njn
00:36pulsebotCheck-in: - Mike Hommey - Bug 1365194 - Remove parts of the hacks for memalign in mozjemalloc. r=njn
00:36pulsebotCheck-in: - Mike Hommey - Bug 1365194 - Compile mozjemalloc as C++. r=njn
00:55pulsebotCheck-in: - Mike Hommey - Bug 1365473 - Rename jemalloc_types.h to mozjemalloc_types.h. r=njn
01:32pulsebotCheck-in: - DimiL - Bug 1364607 - Add a test for empty Safe Browsing updates. r=francois
01:35geraldSometimes an `hg evolve` is stuck doing nothing (after first telling me it's going to move X atop Y). Anyone else seeing that? (mac, hg 4.1.3) I can work around that by doing a manual `hg rebase -d Y -s X`.
01:38pulsebotCheck-in: - Matt Brubeck - servo: Merge #16881 - Fix containing block size for absolute root element (from mbrubeck:absolute_root); r=stshine
01:53pulsebotCheck-in: - Phil Ringnalda - Backed out changeset e53686543601 (bug 1303781) for timing out in browser_popupNotification_no_anchors.js
01:57pulsebotCheck-in: - Lee Salzman - Bug 1364628 - implement thread-safe variants of FT_New_Face/FT_Done_Face. r=jrmuizel
01:57pulsebotCheck-in: - Lee Salzman - Bug 1364628 - give each webrender blob image renderer thread its own thread-local FT_Library to work around unsafe FT_Face rasterization/metrics calls. r=jrmuizel
01:59pulsebotCheck-in: - Thomas Nguyen - Bug 1365466 - Remove enablePrivilege calls in SafeBrowsing test case. r=hchang
02:17pulsebotCheck-in: - Makoto Kato - Bug 1352958 - Use UNIFIED_SOURCES into dom/webbrowserpersist. r=overholt
02:29markhdbaron: thanks for replying to that post. To clarify, you are suggesting that we arrange for the loop to only make progress *inside* the idle-callback, and while timeRemaining is suitably positive?
02:29dbaronmarkh: yeah, I think that's how the API is intended to be used
02:30markhI'm struggling to work out how that loop would be written?
02:30dbaronmarkh: although I guess that doesn't integrate well with promises?
02:30markheg, how that PlacesUtils.jsm function could leverage that
02:31markhI guess a generator of promises could make it work, but that's not how async functions are naturally written
02:31dbaronmarkh: bz and wanderview also replied... and I suspect they're closer to the rIC code than I am.
02:38pulsebotCheck-in: - JW Wang - Bug 1365494 - Remove the call to AbstractThread::GetCurrent() in ScriptLoader.cpp. r=jonco
02:38CJKumattwoodrow: ping
02:38mattwoodrowCJKu: pong
02:39CJKumattwoodrow: hi matt, regarding to your proposal in 1352499
02:39CJKumattwoodrow: I know miko is working on top FrameLayerBuilder from mutating display items
02:39CJKumattwoodrow: is there anything that I can help with?
02:40CJKumattwoodrow: I think some bugs that thinker filed have overlap with your current work
02:41mattwoodrowCJKu: I think the most important thing we could do we be to work on making display building faster, but without using caching (so that it improves first-paint)
02:41mattwoodrowretained-display lists will help later paints a lot, but well still be slow for the first one (and for anything that invalidates a lot)
02:41mattwoodrowDoing profiling of display list building and trying to make that just faster in general would be really valuable
02:42mattwoodrowThings like Bug 1205475 :)
02:43CJKumattwoodrow: ic, profiling and find current bottleneck in buldingdisplaylist
02:46mattwoodrowCJKu: Yeah! FrameLayerBuilder too
02:46CJKumattwoodrow: I found most bullets in your document do not have associate bug(such as Compute correct visible regions. Currently nsDisplayListBuild...)
02:46pulsebotCheck-in: - Alfredo Yang - Bug 1363647 - reduce MIN_SUCCESSIVE_FRAMES to 3 for a very short mp3 file. r=JanH
02:47mattwoodrowCJKu: We havent be using individual bugs, since a lot of the pieces wont be able to land on their own
02:47mattwoodrowWork in happening here -
02:49mattwoodrowCJKu: The other thing would be
02:49CJKumattwoodrow: IC, I will keep my eye on this it.
02:50mattwoodrowI have code for that on the retained-dl repo, but getting that moved onto m-c would help me! And be a good start for bug 1351313
02:51CJKumattwoodrow: ok, I will grep the code and try to understand what you had done
02:52pulsebotCheck-in: - Dzmitry Malyshau - servo: Merge #16891 - Using TransformStyle::Preserve3D (from kvark:preserve3d); r=glennw,emilio
02:52mattwoodrowCJKu: Retain AGRs between paints, and dynamically update them as needed -> Switch the display list builder arena over to nsPresArena since it supports allocating and freeing per-type"
02:55CJKumattwoodrow: roger. thank for your time. Let me survey relative code deeper, will share what my thought after.
02:56mattwoodrowcool, thanks!
03:10pulsebotCheck-in: - JW Wang - Bug 1365504 - Remove the call to AbstractThread::GetCurrent() in ServiceWorkerUpdaterChild.cpp. r=bkelly
03:44pulsebotCheck-in: - cku - Bug 1310885 - Part 11. (gecko) Clean up unused things. r=heycam
03:56pulsebotCheck-in: - JW Wang - Bug 1365520. P1 - remove MDSM::mSentLoadedMetadataEvent and MDSM::EnqueueLoadedMetadataEvent(). r=kaku
03:56pulsebotCheck-in: - JW Wang - Bug 1365520. P2 - remove mMetadataTags for we can pass |aMetadata->mTags| without storing it. r=kaku
03:59pulsebotCheck-in: - Makoto Kato - Bug 1364840 - Part 1. Add sys/auxv.h to system-headers. r=glandium
03:59pulsebotCheck-in: - Makoto Kato - Bug 1364840 - Part 2. Generate aarch64 build rule to add CRC32 feature on aarch64. r=lsalzman
04:02pulsebotCheck-in: - 32 changesets (bugs 1302773, 16910, 1365708, 1302702, 1349155 and 12 other bugs)
04:06pulsebotCheck-in: - 37 changesets - merge mozilla-inbound to mozilla-central a=merge
04:29pulsebotCheck-in: - Chris Manchester - Bug 1307301 - Pack symbols with a python helper and mozjar instead of zip. r=ted
04:29pulsebotCheck-in: - Chris Manchester - Bug 1307301 - Don't attempt to compress compressed files when packing the symbols archive. r=ted
05:06xidornare the tc-M(mdaX) tasks of quantum render target permafailu on try?
05:06xidornit is confusing...
05:07pulsebotCheck-in: - Brian Birtles - Bug 1363107 - Check if the startTime is set before using it in SampleAnimationForEachNode; r=hiro
05:24pulsebotCheck-in: - Masayuki Nakano - Bug 1360135 Get rid of nsPrintData::mSelectedPO because it's unused r=dholbert
05:31pulsebotCheck-in: - Paul Rouget - servo: Merge #16903 - make gl context current on browser initialization (from paulrouget:makeCurrentOnNew); r=emilio
06:15pulsebotCheck-in: - 39 changesets - Merge mozilla-central to autoland
06:16pulsebotCheck-in: - 34 changesets - Merge mozilla-central to mozilla-inbound
06:16pulsebotCheck-in: - JW Wang - Bug 1365517 - Remove the call to AbstractThread::GetCurrent() in TestAsyncReturns.cpp. r=kanru
06:20pulsebotCheck-in: - Alastor Wu - Bug 1364028 - enable specific robocop tests on Autophone. r=bc
06:39pulsebotCheck-in: - Wander Lairson Costa - Bug 1365350: make uploader task a dependency of tests. r=dustin,garndt
06:45pulsebotCheck-in: - Randell Jesup - Bug 1349233: allow SSRC changes in VideoConduits r=bwc a=gchang
06:45pulsebotCheck-in: - Christoph Kerschbaumer - Bug 1359204 - Do not query nested URI within CheckChannel in ContentSecurityManager. r=smaug a=gchang
06:45pulsebotCheck-in: - Christoph Kerschbaumer - Bug 1359204 - Use SystemPrincipal as TriggeringPrincipal when loading page as view-source. r=bz a=gchang
06:45pulsebotCheck-in: - Christoph Kerschbaumer - Bug 1359204 - Test view-source can open link is not blocked by security policies. r=gijs a=gchang
06:45pulsebotCheck-in: - Valentin Gosu - Bug 1359697 - Make CaptivePortalService not poll for status when it is in the NOT_CAPTIVE state r=mcmanus a=lizzard
06:45pulsebotCheck-in: - Boris Zbarsky - Bug 1359859 - Use the right entry global for XBL constructor/destructor/field execution. r=bholley a=gchang
06:45pulsebotCheck-in: - James Cheng - Bug 1362918 - Extend the blacklist to cover certain Galaxy S4 models. r=jolin a=lizzard
06:51pulsebotCheck-in: - Carl Corcoran - Bug 1360493 write a test asserting that Firefox launches without hanging; r=rstrong
06:52KWierso|afkxidorn: certainly looks like some of those chunks are pretty failtastic
06:52pulsebotCheck-in: - Ricky Chien - Bug 1359433 - Fix intermittent browser_net_truncate.js test r=Honza
06:53xidornKWierso|afk: but they are not showing on autoland / m-c
06:53KWierso|afkprobably because they aren't ready to be shown by default, even at tier2
06:53xidornin that case, they shouldn't be shown on try either...
06:53pulsebotCheck-in: - Nils Ohlmeier [:drno] - Bug 1338521: don't over-write remote SSRC with random value r=jesup
06:54xidornit is confusing... I thought there is something wrong in my patch... and it took me quite a while to eventually realize that it is a preexisting permafail :/
06:54pulsebotCheck-in: - Ricky Chien - Bug 1365635 - Fix payload queue and damp regression r=Honza
07:04Tomcat|sheriffdutyjcristau: hi, good morning landed bug 1359697 on release
07:05jcristauTomcat|sheriffduty: thanks, and a good morning to you :)
07:13pulsebotCheck-in: - Carsten "Tomcat" Book - Backed out changeset 420cf7ba089f (bug 1360493) for eslint failure in own test
07:15pulsebotCheck-in: - Evan Tseng - Bug 1365259 - Update the result not found message. r=mconley
07:19pulsebotCheck-in: - cku - Bug 1205475 - Part 1. Hold MAY_HAVE_OPACITY_ANIM/ MAY_HAVE_TRANSFOMR_ANIM information in EffectSet. r=mattwoodrow
07:19pulsebotCheck-in: - cku - Bug 1205475 - Part 2. Prevent multiple EffectSet property look up. r=mattwoodrow
07:19pulsebotCheck-in: - Junior Hsu - Bug 1354405 - Add new telemetry probe for bandwidth cost for rcwn, r=michal, valentin, data-r=bsmedberg
07:19pulsebotCheck-in: - Georg Fritzsche - Bug 1328230 - Store valid process types in a parser-friendly file. r=dexter
07:29Tomcat|sheriffdutybobowen: ping
07:30Tomcat|sheriffdutybobowen: can you take a look at bug 1336657 - you have done a approval request for a obsolete patch, can you check this
07:40bobowenTomcat|sheriffduty: will do
07:40Tomcat|sheriffdutybobowen: thanks! that confused me and gchang a little :)
07:40Tomcat|sheriffdutyso we were feeling we better ask you :)
07:42gchangTomcat|sheriffduty, bobowen : Thanks
07:43bobowenTomcat|sheriffduty, gchang: must have done it on an un-refreshed tab or something
07:44bobowenthanks, bbiab
07:47pulsebotCheck-in: - Paul Rouget - servo: Merge #16904 - Not headless by default (from paulrouget:notHeadLess); r=metajack
08:06pulsebotCheck-in: - Marco Bonardo - Bug 1364002 - Store the user-made search suggestion choice along with userMadeSearchSuggestionChoice. r=past a=gchang
08:14pulsebotCheck-in: - JW Wang - Bug 1365516 - Remove the call to AbstractThread::GetCurrent() in WebRenderImageLayer.cpp. r=sotaro
08:15navidrWhy did firefox switched from Cairo to Skia ? Cairo was that slow ?
08:18freddybnavidr: see thread!topic/
08:18navidrfreddyb: awesome, let me read it.
08:18freddybactually, that does not really contain the "why"
08:19navidrfreddyb: I am interested in "why". I am considering Skia or Cairo for learning. I am some spare time. writing some small graphical 2d game. I am thinking learning which one would be better
08:20navidr it seems skia generally is faster than cairo but in some cases cairo do outperform skia massively
08:30allstarschhsmaug: ping
08:30smaugallstarschh: hi
08:30allstarschhsmaug: hi, quick question, does setTimeout work in srcdoc ?
08:31smaugallstarschh: sure it should
08:32allstarschhsmaug: okay, thanks :), I found a test, looks like the setTimeout is not fired, I'll check more
08:32smaugallstarschh: data:text/html,<iframe srcdoc=&quot;<script>setTimeout(alert(&#39;foo&#39;))</script>&quot;>
08:38allstarschhsmaug: cool, that is a quick answer, BTW, don&#39;t you need </iframe> at the end?
08:39smaugallstarschh: I think parser implicitly adds that when the data contains only <iframe>
08:40allstarschhsmaug: okay, I recall that we need to do <iframe></iframe>, you said parser adds &#39;</iframe>&#39; part ?
08:41smaugwell, I guess it closes unclosed tags somehow at the end of the stream
08:41smaughsivonen would know what it exactly does
08:42allstarschhsmaug: okay
08:42allstarschhsmaug: when I modifed a XUL test, I noticed it uses <iframe .../>
08:42smaugallstarschh: ahh, XUL is XML
08:42smaugI used HTML
08:42allstarschhsmaug: ah, I see
08:42allstarschhsmaug: thanks :)
08:59pulsebotCheck-in: - Mark Banner - Bug 1361760 - Enable some Mozilla ESLint recommended rules for dom/indexedDB/test/unit. r=bevis
08:59pulsebotCheck-in: - Mark Banner - Bug 1361760 - Enable some Mozilla ESLint recommended rules for dom/indexedDB/test/unit - automatic fixes. r=bevis
08:59pulsebotCheck-in: - Mark Banner - Bug 1361760 - Enable some Mozilla ESLint recommended rules for dom/indexedDB/test/unit - manual fixes. r=bevis
09:03wesley_huangdaleharvey: ping
09:03daleharveywesley_huang: pong
09:04wesley_huangwe&#39;re gonna triage
09:04wesley_huangfor PWA
09:04wesley_huangroom = indigo
09:56pulsebotCheck-in: - Andrea Marchesini - Bug 1365598 - Updating the Constructor of HTMLOptionElement, r=smaug
09:57pulsebotCheck-in: - sole - Bug 1365216 - DevTools: bring over documentation for how to find documentation into the tree. r=pbro
09:57pulsebotCheck-in: - sole - Bug 1365216 - Follow up. r=pbro
10:11pulsebotCheck-in: - Marco Bonardo - Bug 1364488 - Recover from a Talos regression due to recent changes to the Star button. r=standard8
10:11pulsebotCheck-in: - Marco Bonardo - Bug 1364488 - Allow fetch to use a concurrent connection. r=standard8
10:43pulsebotCheck-in: - Shih-Chiang Chien - Bug 1365566 - use PR_GetSystemInfo to get OS version on Firefox desktop. r=dragana.
11:00pulsebotCheck-in: - Masatoshi Kimura - Bug 1363055 - Find a nearest screen if no screen overlaps. r=kanru
11:03pulsebotCheck-in: - Shawn Huang - Bug 1365478 - Use BoolVarCache to cache preferences in XMLHttpRequest, r=baku
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 1: Add flag to nsISHEntry to indicate if it was originally loaded in this process. r=bz
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 2: Limit the http pages that will load in the file content process with allowLinkedWebInFileUriProcess pref. r=Gijs, r=mystor
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 2.5: Move logic in the parent for switching in and out of Large Allocation process into browser.js RedirectLoad. r=mystor
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 3: In tabbrowser.xml adoptTab, make sure the new tab is in the same process as adoptee. r=Gijs
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 4: Select correct remote type upfront when passed window arguments in browser.js onDCL. r=Gijs
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 5: Enable pref allowLinkedWebInFileUriProcess to allow related http(s) content to top level load in file content process. r=Gijs
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 6 prologue: Allow BrowserTestUtils.waitForNewTab to optionally wait for the page in the new tab to load. r=Gijs
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 6: Include more tests cases for allowLinkedWebInFileUriProcess pref. r=Gijs
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 7: Check that we can form post from file:// URL to http:// URL. r=bz
11:09pulsebotCheck-in: - Bob Owen - Bug 1351358 Part 8: Check that reload doesn&#39;t cause new process when large allocation browser in tab group. r=mystor
11:11pulsebotCheck-in: - Gijs Kruitbosch - Bug 1365294 - swap position of the fixed overflow list, r=mikedeboer
11:17pulsebotCheck-in: - Gerald Squelart - Bug 1365534 - Remove unneeded IsMainThread assertions - r=cpearce
11:17pulsebotCheck-in: - Gerald Squelart - Bug 1365534 - In non-e10s, open temp file in FileBlockCache thread - r=cpearce
11:17pulsebotCheck-in: - Gerald Squelart - Bug 1365534 - Close temp file in FileBlockCache thread - r=cpearce
11:23pulsebotCheck-in: - Do Gottwald - Bug 1365683 - Increase horizontal margin around the location bar and the search bar. r=dale
11:36pulsebotCheck-in: - cnevinc - Bug 1358946 - Strip about:reader in intent data uri. r=maliu
11:54pulsebotCheck-in: - Sebastian Hengst - Backed out changeset 86af05828caf (bug 1358946) for Android bustage in r=backout
12:18pulsebotCheck-in: - 39 changesets (bugs 1365788, 1361206, 1303781, 817374, 1365420 and 23 other bugs)
12:19pulsebotCheck-in: - 16 changesets - merge mozilla-inbound to mozilla-central a=merge
12:19pulsebotCheck-in: - Timothy Guan-tin Chien - Bug 1365869 - Bug 1365869 - Better contrast comparison for Selection#setColors(). r=masayuki
12:20pulsebotCheck-in: - Petr Sumbera - Bug 1365620 - Option -rpath-link is not supported by Solaris linker. r=glandium
12:20pulsebotCheck-in: - Petr Sumbera - Bug 1364066 - Re-enable bundled NSS on Solaris. r=glandium
12:24pulsebotCheck-in: - Yury Delendik - Bug 1362084 - Add binary source and JS source maps support. r=luke
12:32pulsebotCheck-in: - Bob Owen - Bug 1175267 Part 1: Handle file:// URI xpi files in chrome code directly instead of routing through content browser. r=mossop
12:32pulsebotCheck-in: - Bob Owen - Bug 1175267 Part 2: Ensure that process does not switch for file:// XPI installs. r=mossop
12:36pulsebotCheck-in: - Andreas Pehrson - Bug 1363253 - Ignore events specific to preview window also when decrementing. r=florian, a=test-only
12:36pulsebotCheck-in: - Neil Deakin - Bug 1358713 - When a popup&#39;s anchor does not have a frame originally, don&#39;t have the popup update its position to follow it. r=tn, a=gchang
12:36pulsebotCheck-in: - Kevin Chen - Bug 1363677 - Skip Flush() and EndDraw() if the ID2D1DeviceContext is stale. r=dvander, a=gchang
12:36pulsebotCheck-in: - Bob Owen - Bug 1336657 - On Windows add a sandbox policy rule to allow read / write access to content temp dir. r=jimm, a=gchang
12:36pulsebotCheck-in: - Felipe Gomes - Bug 1363083 - Remove TabMixPlus and Mega from e10s blocklist. r=mrbkap, a=gchang
12:36pulsebotCheck-in: - Julien Cristau - Bug 1358420 - Use staged rollout for fennec beta builds. r=aki, r=jlorenzo, a=gchang
12:46pulsebotCheck-in: - Do Gottwald - Bug 1365846 - Use context-fill for identity, connection and tracking protection icons in the site identity block. r=johannh
12:49pulsebotCheck-in: - Boris Zbarsky - Bug 1359859 - Use the right entry global for XBL constructor/destructor/field execution. r=bholley, a=gchang
12:49pulsebotCheck-in: - Ryan VanderMeulen - Bug 1300311 - Fuzz some reftests on WinXP. a=test-only
12:56smaugif I have a try run without talos runs and would like to see also talos results, how do I talos?
12:56smaugor do I need to re-push?
12:56smaugTomcat|sheriffduty: you might know
13:00pulsebotCheck-in: - 41 changesets - Merge mozilla-central to mozilla-inbound
13:01pulsebotCheck-in: - 17 changesets - Merge mozilla-central to autoland
13:01Tomcat|sheriffdutysmaug: not a talos expert but jmaher|afk might know
13:01RyanVMsmaug: you should be able to add them from Treeherder&#39;s UI assuming you pushed on a recent parent rev
13:02smaugI re-pushed to try
13:02smaugRyanVM: I couldn&#39;t figure out how
13:02RyanVMare you familiar with the Add New Jobs option?
13:02RyanVMon the right hand side, there&#39;s a drop marker
13:02smaugI saw something like that and clicked it :)
13:03RyanVMthere&#39;s actually a Trigger all Talos jobs option there too, but I must confess to never having used it
13:03smaughmm, where is that
13:03RyanVMhmm, I see it on your push
13:04smaugyes, that one
13:04RyanVMbut I also am part of the sheriffing group, which has more superpowers :P
13:04smaugso I did click Add New Jobs
13:04smaugbut I don&#39;t have that anymore visible
13:04smaugI guess something changed
13:04smaugand I&#39;m supposed to do now something
13:22Tomcat|sheriffdutykats emilio you around ?
13:22emilioTomcat|sheriffduty: yes
13:22katsTomcat: yeah
13:23Tomcat|sheriffdutyhey emilio kats do you know of a graphics change that could turn our win7 hardware into bluescreen
13:23Tomcat|sheriffdutylike driver test or so
13:23emilioTomcat|sheriffduty: I don&#39;t :/
13:23Tomcat|sheriffdutykats emilio could you join #buildduty for a moment they are trying to investigate this
13:27atoHow would I get a <xul:browser>s <xul:tab>?
13:28atoOn a tab I can call linkedBrowser to get the browser, but not sure how to do the reverse.
13:28atoThe use case I have is that I need to find out that a TabOpen event is on a tabbed browser, e.g. now a frame.
13:29Tomcat|sheriffdutyBas: ping
13:29Gijsato: TabOpen events only get fired for tabs
13:30BasTomcat|sheriffduty: pong
13:30atoGijs: Ooh, so my check is completely unnecessary.
13:30Tomcat|sheriffdutyBas: can you join #buildduty
13:30Gijsato: gBrowser.getTabForBrowser(browser) would normally work for your original question, but if it&#39;s really just for TabOpen events, those get fired by the tabbrowser
13:30Tomcat|sheriffdutyseems one of your try runs affect win7 masschines
13:30Gijsato: I think so
13:30pulsebotCheck-in: - Henry Chang - Bug 1364857 - Reject pending promises for actor when it&#39;s going to be destroyed. r=kanru
13:31atoGijs: Im porting some code that uses loadFrameScript recursively to register new browsers from an IPC message to looking at the TabOpen event, but of course if that only fires on tabs, then my check is unnecessary.
13:31Gijsato: the original target of that event should also be the tab in question
13:31atoWell that is excellent!
13:32Gijsato: I&#39;m not sure I understand the loadFrameScript stuff... but you can pass true as the second param to loadFrameScript on a group message manager, and then the framescript will load into new tabs automatically
13:32Gijsin case that&#39;s helpful...
13:33atoGijs: Yes, we do that and we want to get rid of it because the browsers are sometimes blown away by remoteness changes.
13:33atoGijs: My working theory is that tracking top-level browsing contexts (i.e. tabbed browsers) via a combination of browser-delayed-startup-finished, Browser:Init, and TabOpen is a better approach.
13:35Gijsato: hm, depends what you&#39;re trying to do, I guess :)
13:35nemoyou know, whenever someone posts a neat writeup like this
13:36nemoI really wish they&#39;d offer a link to the original png too
13:36nemo&#39;cause those jpeg results seem way too terrible to be plausible
13:36nemowould like to feed into mozjpeg for example
13:36atoGijs: Im trying to keep a cache of all current tabbed browsers, like tabbrowsers _outerWindowIDBrowserMap, but for all ChromeWindows and using permanentKey instead of outerWindowID.
13:36pulsebotCheck-in: - Gregory Arndt - Bug 1364266 - Specify task priority within task definition r=dustin
13:39Gijsato: what for? :)
13:39Gijsato: what&#39;s using the cache, I mean
13:41atoGijs: When a new tabbed browser appears, Marionette assigns a UUID to it because the permanentKey isnt serialisable and we need a way to communicate the identity of a tab to an RPC client so it can later be located.
13:43atoSo to not have to search through all tabs in all tabbrowsers on all windows every time we have to, say, access a browsers message manager, we can look it up in the UUID (permanentKey) to browser cache.
13:48Gijsato: OK. If this is based on the permanentKey then yeah, I think using TabOpen is sensible. You might need to check how to deal with tabs moving between windows
13:48* Gijs isn&#39;t really sure what events would fire for that and if that works for your usecase
13:49atoGijs: As far as I understand that shouldnt change the permanentKey, and I dont think I need to keep track of exactly which window a tab lives in.
13:49atoGijs: E.g. its flattening all tabbed browsers from all windows into one iterator you could say.
13:49Gijsato: not the currentKey, but the tab element I think changes - though I could be wrong...
13:49* Gijs hasn&#39;t looked at that code in a while
13:49Gijserr, s/currentKey/permanentKey/
13:50Gijsato: anyway, just wanted to flag that up as something to test, is all :)
13:50atoYeah, one worry I have is that I dont currently track when ChromeWindows close and leave a bunch of event listeners hanging around.
13:50atoYes, this is (again) very useful (-:
13:50atoI see that tabbrowser.xml deals with most of this so Im trying to follow that as closely as I can.
13:51atoExcept with the complexity that I need to do this across _all_ tabbrowsers.
14:00chuttenI don&#39;t suppose anyone listening here knows what XULStore is?
14:12pulsebotCheck-in: - Nicolas B. Pierron - Bug 1364118 - List JS bytecode cache preferences in all.js. r=mrbkap
14:17pulsebotCheck-in: - Jan Odvarko - Bug 1364386 - Use only file name for search; r=nchevobbe
14:17pulsebotCheck-in: - Jan Odvarko - Bug 1364386 - Update tests; r=nchevobbe
14:24pulsebotCheck-in: - Anthony Ramine - servo: Merge #16889 - Refactor how calc() clamping is done on computed values (fixes #15296) (from nox:clamp-calc); r=stshine
14:26pulsebotCheck-in: - Tom Tromey - Bug 1317962 - use source maps in stack traces in the console; r=jryans
14:31atoGijs: So TabClose does not fire if you close the tabbrowser/window they are in?
14:46Gijsato: it does not. unload will fire on the window.
14:47Gijsato: there&#39;s a bug on file for changing this, I think, but it might have been wontfixed, I don&#39;t remember.
14:47atoListening for the window unload event will require a frame script, but the frame script might be lost at any point in time due to a remoteness change.
14:47Gijsjwatt: uh, so, what&#39;s up with bug 1362120 ?
14:48Gijsjwatt: it looks like it landed, got backed out, and then the reporter came along and closed it as invalid, so now I&#39;m just confused
14:48atoGijs: tabbrowser.xml obviously overcomes this because every tab close/window close action is explicit, e.g. clicking to close a tab.
14:48Gijsato: no no, I mean the parent window
14:48atoGijs: The ChromeWindow?
14:48jwattGijs: you&#39;re not the only one
14:49Gijsjwatt: he did this to several bugs...
14:49Gijsjwatt: should we just reopen them? :s
14:49Gijs*he or she
14:49jwattGijs: I don&#39;t know about the others, but I&#39;ll test that one and most likely reopen
14:51Gijsjwatt: thanks!
14:51jwattGijs:weird behavior though
14:51jwattmaybe they decided they didn&#39;t want bugmail?
14:51jwattwho knows
14:53Gijshm, I think I was confused about them doing this multiple times by my bugmail reading
14:53* Gijs can&#39;t find any traces of it in bugzilla
14:53Gijsthis is what happens when I try too much bugmail at once :s
14:53Gijs*to read too much bugmail at once
14:53* Gijs takes a break
14:57Tomcat|sheriffdutyemilio: we need a test update again for and
14:59pulsebotCheck-in: - ffxbld - No bug, Automated HSTS preload list update from host bld-linux64-spot-363 - a=hsts-update
14:59pulsebotCheck-in: - ffxbld - No bug, Automated HPKP preload list update from host bld-linux64-spot-363 - a=hpkp-update
15:00emilioTomcat|sheriffduty: looking, thanks :)
15:01pulsebotCheck-in: - Michelangelo De Simone - Bug 1357911 - Baldr: update names section format (r=luke)
15:02pulsebotCheck-in: - ffxbld - No bug, Automated HSTS preload list update from host bld-linux64-spot-361 - a=hsts-update
15:02pulsebotCheck-in: - ffxbld - No bug, Automated HPKP preload list update from host bld-linux64-spot-361 - a=hpkp-update
15:03pulsebotCheck-in: - ffxbld - No bug, Automated HSTS preload list update from host bld-linux64-spot-327 - a=hsts-update
15:03pulsebotCheck-in: - ffxbld - No bug, Automated HPKP preload list update from host bld-linux64-spot-327 - a=hpkp-update
15:06mystorsewardj: ping re bug 1365309
15:10pulsebotCheck-in: - Rob Wu - Bug 1299053 - Remove unsupported &quot;notification&quot; type. r=aswan
15:10pulsebotCheck-in: - Rob Wu - Bug 1299053 - Refactor test/browser/browser_ext_getViews r=aswan
15:10pulsebotCheck-in: https: