The first official 20171104 builds are out

Discussion about official Mozilla Firefox builds
Josa
Posts: 7360
Joined: July 28th, 2009, 4:52 pm

The first official 20171104 builds are out

Post by Josa »

Previous Nightly Builds thread 20171103

The first official 20171104 NIGHTLY builds are out
.The Official Firefox 57 RC1 builds are not yet out

Previous NIGHTLY: 20171103 (Fx 58.0a1)
Previous BETA: 20171103 (Fx 57.0b14)

Mozilla Inbound Builds
Treeherder
Release Notes: NIGHTLY | BETA
Changelogs: NIGHTLY | BETA


Fixed:
  1. NIGHTLY +75
  2. #1412643
  3. #1395922 [Core:Audio/Video: Playback]-Handle HTMLMediaElement::SetMediaKeys asynchronously. [Uns][]
  4. #1411808 [Core:Audio/Video: Playback]-Run MediaCache::Update() off the main thread [Uns][]
  5. #1413484 [Core:Audio/Video: Playback]-Tighten up assertions in ChannelMediaResource [Uns][]
  6. #1413734 [Core:Audio/Video: Playback]-update av1 decoder [Uns][]
  7. #1374247 [Core:CSS Parsing and Computation]-stylo: We don't implement the hack to allow child combinators to match across XBL <children> elements (could we remove it?). [Uns][]
  8. #1403814 [Core:DOM: Security]-Block toplevel data: URI navigations only if openend in the browser [Uns][[tor][domsecurity-active]]
  9. #1403870 [Core:DOM: Security]-Potentially allow navigations to data:application/json [Uns][[domsecurity-active]]
  10. #1412824 [Core:DOM: Security]-Refactor MaybeCloseWindow to allow closing the window if toplevel data URI navigation is blocked [Uns][[domsecurity-active]]
  11. #1412345 [Core:DOM]-Crash in ExpandedPrincipal::GetHashValue, with MOZ_CRASH(extended principal should never be used as key in a hash map) [Win][]
  12. #1414359 [Core:DOM]-Remove "creator" bits from WebIDL parser [Uns][]
  13. #1410159 [Core:Disability Access APIs]-Add more methods to HandlerProvider::GetEffectiveOutParamIid [Win][aes+]
  14. #1403358 [Core:Disability Access APIs]-Disable Accessibility broken by ipc/mscom/ProxyStream.cpp:312:5: error: ‘ActivationContext’ has not been declared [Uns][[tor]]
  15. #1408532 [Core:Graphics: Layers]-Crash in shutdownhang | mozilla::layers::CompositorThreadHolder::Shutdown [Win][[gfx-noted]]
  16. #1414097 [Core:Graphics: WebRender]-Convert nsDisplaySelectionOverlay to webrender [All][[wr-reserve]]
  17. #1412848 [Core:Graphics: WebRender]-Crash in memcpy | mozilla::gfx::RecordedScaledFontCreationByIndex::Record<T> [Win][[wr-mvp] [gfx-noted]]
  18. #1413397 [Core:Graphics: WebRender]-Empty border should not fall back [Uns][[wr-mvp]]
  19. #1408421 [Core:Graphics: WebRender]-RenderBackend thread does not overlap cpu time with the Render thread [Uns][[wr-mvp]]
  20. #1413178 [Core:Graphics: WebRender]-Update webrender to fae962bfd6e1997f4b921ee93c3c1cc5abca3256 [Uns][[wr-mvp] [gfx-noted]]
  21. #1414036 [Core:Graphics: WebRender]-regression: BMO's buttons don't change when hovering over them [All][[wr-mvp]]
  22. #1413680 [Core:Graphics: WebRender]-regression: blob-images: svg invisible on web.whatsapp.com [Lin][[wr-mvp]]
  23. #1390741 [Core:Graphics: WebRender]-webrender: panels of remote webextensions are white [Win][[wr-mvp][gfx-noted]]
  24. #1410736 [Core:Internationalization]-Remove remaining uses of `general.useragent.locale` [Uns][]
  25. #1414186 [Core:Internationalization]-Use GetUserPreferredUILanguages for OSPreferences::SystemLocales [Uns][]
  26. #1413944 [Core:JavaScript Engine: JIT]-Remove unreachable code in VMFunction's StringFromCharCode [Uns][]
  27. #1413907 [Core:JavaScript Engine]-Change InitPropertyOperation to accept a PropertyName directly [Uns][]
  28. #1409187 [Core:JavaScript Engine]-Remove __proto__ setter telemetry [Uns][[js:correctness][js:tech-debt]]
  29. #1386511 [Core:JavaScript: GC]-Improve the formatting of GC markers [Uns][]
  30. #1376891 [Core:JavaScript: GC]-Investigate eagerly collecting the nursery in an idle callback [Uns][[qf:p3]]
  31. #1413063 [Core:JavaScript: GC]-UBSan: js/src/gc/Nursery.cpp:486:20: runtime error: division by zero [@ calcPromotionRate] [Uns][]
  32. #1386146 [Core:JavaScript: Internationalization API]-Add support for hourCycle to Intl.DateTimeFormat [Uns][]
  33. #1406993 [Core:JavaScript: Standard Library]-Pick newer equivalent years when the year >= 2038 [Uns][]
  34. #1119777 [Core:JavaScript: Standard Library]-Remove non-standard Function.prototype.isGenerator [All][]
  35. #1410292 [Core:Layout: Form Controls]-[DateTimeInput] TypeError: this.mInputElement is null [Uns][]
  36. #1413622 [Core:Networking: Cache]-UBSan: netwerk/cache/nsCacheService.cpp:3067:63: division by zero [@ nsCacheService::LogCacheStatistics] [Uns][[necko-triaged]]
  37. #1414046 [Core:Networking: Cache]-UBSan: null pointer passed as argument declared to never be null [@ mozilla::net::CacheFileMetadata::WriteMetadata] [Uns][[necko-triaged]]
  38. #1413845 [Core:Plug-ins]-PluginPRLibrary looks unused [Uns][]
  39. #1413400 [Core:Preferences: Backend]-Avoid manual refcounting in libpref [Uns][]
  40. #1413413 [Core:Preferences: Backend]-Remove support for extensions having their own prefs file [Uns][]
  41. #1409259 [Core:Security: PSM]-Implement a console warning for Symantec CAs affected by the distrust plan [All][[psm-assigned]]
  42. #1406446 [Core:Widget]-Expose EventStateManager::IsHandlingUserInput() (or equivalent) in InputContextAction [Uns][]
  43. #1414461 [Core:XPConnect]-Audit Xray code for OOM problems with ensureHolder() [Uns][]
  44. #1413550 [Firefox:Activity Streams: Newtab]-Add per-platform styles, per-locale html and bug fixes to Activity Stream [Uns][]
  45. #1412930 [Firefox:Activity Streams: Newtab]-Package localized prerendered files [Uns][]
  46. #1410940 [Firefox:Bookmarks & History]-Copy & Pasting multiple items into the Library window gets the insertion indexes wrong [Uns][[fxsearch]]
  47. #1404850 [Firefox:Developer Tools: Console]-Fix self-XSS protection in new console and enable test [Uns][]
  48. #1412273 [Firefox:Developer Tools: Inspector]-CSS shapes editor swatch icon invisible in dark theme [Uns][[designer-tools]]
  49. #1282721 [Firefox:Developer Tools: Inspector]-Make the global shape in the CSS Shapes highlighter editable [Uns][[designer-tools]]
  50. #1413605 [Firefox:Developer Tools: Inspector]-MarkupContainer creates event listeners on window for each node in the Inspector [Uns][]
  51. #1413941 [Firefox:Developer Tools: Netmonitor]-Fix broken netmonitor launchpad regression from bug 1360457 [All][]
  52. #1407561 [Firefox:Developer Tools: Netmonitor]-Lazy loading of tooltip text when user onhover the status column [Uns][[good first bug][mentor-lang=zh][lang=js]]
  53. #1413829 [Firefox:Developer Tools: Netmonitor]-RequestListContent scroll to bottom should avoid sync reflow [All][]
  54. #1408964 [Firefox:Developer Tools: Netmonitor]-The Network panel should be auto resumed on reload [Uns][]
  55. #1412359 [Firefox:Developer Tools: Responsive Design Mode]-Data confusion when removing custom devices [All][[designer-tools]]
  56. #1411906 [Firefox:Developer Tools]-[dt-onboarding] Add close button to devtools onboarding page [Uns][]
  57. #366192 [Firefox:General]-Disallow htppfoobor schemes when only http/https are valid [All][]
  58. #1413296 [Firefox:General]-Remove e10sCohort annotation from Telemetry environment [All][]
  59. #1373258 [Firefox:New Tab Page]-Convert PageThumbsProtocol.js to PageThumbsProtocol.cpp [Uns][]
  60. #1368209 [Firefox:Sync]-Break up `Engine::_processIncoming` [Uns][]
  61. #1388224 [Firefox:Sync]-Intermittent services/sync/tests/unit/test_service_getStorageInfo.js | test_invalid_json - [test_invalid_json : 70] "undefined" == 500 | test_invalid_json - [test_invalid_json : 82] "NS_ERROR_NOT_AVAILABLE" == "SyntaxError" [Uns][]
  62. #1413685 [Firefox:Sync]-Unify AsyncResource and Resource in sync resource.js. [Uns][]
  63. #1366067 [Firefox:Sync]-Use JSONFile.jsm to write Sync metadata files [Uns][]
  64. #1414216 [Firefox:Tabbed Browser]-Disable "tabsintitlebar" when disabled by -moz-gtk-csd-available media feature [Uns][]
  65. #1413499 [Firefox:Theme]-Simplify disabled toolbar button styling [All][[reserve-photon-visual]]
  66. #1393627 [Firefox:Tracking Protection]-Expose the full TP UI preferences [Uns][]
  67. #1413322 [Toolkit:Add-ons Manager]-Stop loading old-style langpacks [Uns][]
  68. #1333126 [Toolkit:Crash Reporting]-Add support for CFI-based stack-walking to the minidump-analyzer tool [Uns][]
  69. #1413110 [Toolkit:Form Manager]-[Form Autofill] No need to compute fields on tombstone while data migration [All][[form autofill]]
  70. #1411640 [Toolkit:Themes]-Consolidate <radio> bindings across platforms [Uns][[xbl-flatten-inheritance]]
  71. #1414237 [Toolkit:Themes]-Remove bogus styles from listbox.css on Linux [Lin][]
  72. #1413979 [Toolkit:Themes]-Remove bogus styles from tree.css on Linux [Lin][]
  73. #1412361 [Toolkit:Themes]-Remove checkbox-baseline and checkbox-radio xbl bindings [Uns][[xbl-remove-unused]]
  74. #1347190 [Toolkit:WebExtensions: Frontend]-Add Google Chrome border/ separator color properties [Uns][triaged]
  75. #1383300 [Toolkit:WebPayments UI]-Show origin and total roughly similar to the UX specs [Uns][]
  76. #1403231 [Toolkit:XUL Widgets]-Prototype removal of "image" XBL binding [Uns][[xbl-special-cases]]
  1. Partial Landings/Diagnostic Patches:
    None
  1. Beta 14 -> RC1 Changelog +0
    None

Nightly 58 fixes since 20170921 (Gecko 57) ~2325
Beta 57 fixes since 20170802 (Gecko 56) ~3567

Nightly Blog
Release Calendar
Release Tracking
MozRegression Guide
Last edited by Josa on November 4th, 2017, 10:06 am, edited 1 time in total.
koboltzz
Posts: 36
Joined: December 22nd, 2015, 3:17 am

Re: The first official 20171104 builds are not yet out

Post by koboltzz »

Someone notices that every known browser benchmark (like speedometer 2.0, motionmark, basemark, jetstream, XPRT etc.) show greatly reduced scores?
Previously I get ~220 points in motionmark and now 35 :D
Things I've changed:
layout.css.servo.chrome.enabled, true
layout.display-list.retain, true

BUT they do not really run slower (subjectively equal or faster), it seems that mozilla invented some fast code, that current benchmarks cannot gauge ;)
User avatar
sciguyryan
Folder@Home
Posts: 2181
Joined: November 10th, 2004, 1:33 pm
Location: Wales

Re: The first official 20171104 builds are not yet out

Post by sciguyryan »

koboltzz wrote:BUT they do not really run slower (subjectively equal or faster), it seems that mozilla invented some fast code, that current benchmarks cannot gauge ;)
Unlikely since the code is available for all too see and not too difficult to follow. This isn't really the place for conspiracy theories, though if you think there is an actual bug at play...
Cheers!

Ryan Jones
koboltzz
Posts: 36
Joined: December 22nd, 2015, 3:17 am

Re: The first official 20171104 builds are not yet out

Post by koboltzz »

This was just an information about something I observe and a little joke marked with a ' ;) '. No reason to tell me for what this place is for.
User avatar
streetwolf
Posts: 2700
Joined: August 21st, 2011, 8:07 am
Location: NJ (USA)

Re: The first official 20171104 builds are not yet out

Post by streetwolf »

koboltzz wrote:This was just an information about something I observe and a little joke marked with a ' ;) '. No reason to tell me for what this place is for.
What you might be seeing are add-ons causing this. uBlock Origin results in a lower Speedometer score the last time I checked. I notified Raymond about it and either he did something about it or is waiting on Mozilla. I know a few other add-ons also lowers my Speedometer score. Try running Speedometer with no add-ons.
Intel i9-13900K | ASUS ROG MAXIMUS Z790 HERO DDR5 | 64GB CORSAIR VENGEANCE DDR5 @ 6400 Mhz.
H100i ELITE CAPELLIX XT Liquid CPU Cooler | PNY 12GB GeForce RTX 3080 Ti | 2 CORSAIR 2TB MP600 PRO XT GEN 4
HX1200 PLATINUM PSU | XENEON 32" IPS UHD 144Hz | BenQ 32" UHD | MS Windows 11 Pro
koboltzz
Posts: 36
Joined: December 22nd, 2015, 3:17 am

Re: The first official 20171104 builds are not yet out

Post by koboltzz »

You are right at speedometer :) But motionmark is still down to 35 from 220 points since turning the two mentioned prefs to true. As I said, the curious thing about that is you won't notice such a big performance hit. They are just numbers and they are not important for me.... but it's strange.
User avatar
Edds
Posts: 332
Joined: July 21st, 2010, 1:13 pm
Location: SVK

Re: The first official 20171104 builds are not yet out

Post by Edds »

interesting move:
#1413295 [Firefox:Installer]-Rename "Mozilla Firefox" to "Firefox" on Windows [Uns][[fce-active]]
Laptop: Intel Core i5 1240P | 16GB DDR4 | 500GB NVMe | 1080p | Win 11
Phone: Samsung S24 Ultra | Android 14
Mouse5
Posts: 1279
Joined: April 11th, 2014, 7:34 pm
Location: Sydney Australia

Re: The first official 20171104 builds are not yet out

Post by Mouse5 »

Edds wrote:interesting move:
#1413295 [Firefox:Installer]-Rename "Mozilla Firefox" to "Firefox" on Windows [Uns][[fce-active]]
i dont think its " interesting " i mean we prolly all saw this move ages ago, but eventually it'll move to MACOSX an Linux builds. the devs are prolly using Windows as a Testing Ground before they move the renaming elsewhere.
User avatar
Omega X
Posts: 8225
Joined: October 18th, 2007, 2:38 pm
Location: A Parallel Dimension...

Re: The first official 20171104 builds are not yet out

Post by Omega X »

koboltzz wrote:You are right at speedometer :) But motionmark is still down to 35 from 220 points since turning the two mentioned prefs to true. As I said, the curious thing about that is you won't notice such a big performance hit. They are just numbers and they are not important for me.... but it's strange.
My score is 110 on that site with those enabled and 179 in Fx56. Although I don't see any downsides during normal browsing. There's some sort of regression, you should file a bug.


EDIT:
I ran a few tests:

BTW layout.css.servo.chrome.enabled is not enabled by default due to some issues.

[1] Stylo Content ONLY:: 201
[2] Stylo Chrome ONLY: 124
[3] Stylo Chrome + Content: 174
[4] Stylo + Retain: 110
[5] Stylo + OMTP OFF: 167
[6] ALL ENHANCEMENTS OFF: 134
[7] Fx 56: 179

Conclusion:

- Stylo has a positive effect on the test.
- OMTP has a positive effect on the test.
- Retain is probably doing something it shouldn't.
- Stylo in Chrome has some kind of weirdness that affects the test.

With everything off, it should have mimicked Fx56. But for some reason it didn't. Addons may be to blame but I have not tested that theory yet.

The test also has this issue where if you change tabs after the test is done, the score seems to adjust itself again. Weirdness.
koboltzz
Posts: 36
Joined: December 22nd, 2015, 3:17 am

Re: The first official 20171104 builds are not yet out

Post by koboltzz »

Omega X wrote:... Weirdness.
Thank you for confirming.
User avatar
sciguyryan
Folder@Home
Posts: 2181
Joined: November 10th, 2004, 1:33 pm
Location: Wales

Re: The first official 20171104 builds are not yet out

Post by sciguyryan »

Omega X wrote:- Retain is probably doing something it shouldn't.
It would be interesting to see how this behaves with WebRender enabled too. It may behave more correctly in that circumstance.
Omega X wrote:- Stylo in Chrome has some kind of weirdness that affects the test.
It seems that there is some oddness going on with the way selectors at the "window" scope are affecting web pages - primarily I suspect for OS specific rendering for the likes of form controls and so on. I haven't tested this but it is a possibility.
Cheers!

Ryan Jones
User avatar
Omega X
Posts: 8225
Joined: October 18th, 2007, 2:38 pm
Location: A Parallel Dimension...

Re: The first official 20171104 builds are not yet out

Post by Omega X »

Can someone test to see if YouTube VP9 4K30p content drops frames? Fx56 isn't dropping frames like Nightly.
User avatar
GHM113
Posts: 707
Joined: December 16th, 2015, 3:59 am
Location: Moscow, Russia

Re: The first official 20171104 builds are not yet out

Post by GHM113 »

Nightly 04-11-17-2 crashes when I try to launch the profile manager using "nightly.exe -p".
Sorry for my poor English.
TheVisitor
Posts: 5469
Joined: May 13th, 2012, 10:43 am

Re: The first official 20171104 builds are not yet out

Post by TheVisitor »

GHM113 wrote:Nightly 04-11-17-2 crashes when I try to launch the profile manager using "nightly.exe -p".
Have not tested, but are you sure 'nightly.exe' is a valid file ?

I thought the command line to launch ProfileManager was firefox.exe -p
User avatar
GHM113
Posts: 707
Joined: December 16th, 2015, 3:59 am
Location: Moscow, Russia

Re: The first official 20171104 builds are not yet out

Post by GHM113 »

@TheVisitor you are right, I am trying to launch Nightly/firefox.exe -p but Nightly crashes.
Sorry for my poor English.
Locked