firstname.lastname@example.org last week
Reviewed: - bug 1317674, bug 1295937, bug 1286934, bug 1314056, bug 1285752, bug 1317555, bug 1318182, bug 1319166, bug 1317778, bug 1316724, bug 1263595, bug 1313808, bug 1200951, bug 1318494, bug 1306327, bug 1306329, bug 1318905, bug 1318189, bug 1287048, bug 1313295, bug 1280683, bug 1318247, bug 1320621.
- bug 1311039: Properly detect the default malloc zone on OSX 10.12
- bug 1319345: Change how application.ini and application.ini.h are generated.
- bug 1319350: Add dedicated targets for some one-offs in the build
- bug 1319351: Avoid splash.rc depending on files in dist/branding
- bug 1318352: Allow to run AccEventGen.py and xpild/header.py from any directory
- bug 1319604: cc1plus: warning: unrecognized command line option '-Wno-null-conversion'
- bug 1319637: Transform ObjDirFiles into FinalTargetFiles
- bug 1319959: Can't build because of missing include in security/nss/lib/base/nssbaset.h
- bug 1320006: Android local build failure after bug 1319345
- bug 1317504: Sort things out wrt LD/LINK in configure
- bug 1316957: Include windows.configure conditionally instead of using @depends_win
- bug 1321065: Default to --enable-profiling for nightly milestones.
Other: - git-cinnabar 0.4.0rc
- More of the same
email@example.com last week
GCC 6 fixes:
- Pushed on with removal of ZEROING_NEW (bug 1232696). Got an initial patch to test.
- Iterated bug 1316556 (Remove zeroing allocation in class nsIPresShell).
- Landed push-to-valgrind-mochitest finally (yay!) (bug 1232696). Try syntax is "-b o -p linux64 -u mochitest-valgrind -t none".
- Worked to greenify the mochitest-valgrind runs. Landed bug 1248365, giving about 21 out of 40 green. Harvested two more problems, bug 1320894 and bug 1320905. Fixing them both should get to about 28 out of 40 green.
- Used the mochitest-valgrind runs for real to search for problems with the gcc 6 porting work.
- Ran Gtests on Valgrind, harvested bug 1318012, bug 1318030, got them fixed.
- Found and fixed a quadratic-slowdown problem in V shown by the Gtest runs.
- Observed that bug 1309851 (short-run Valgrind automation timeout probs) isn't entirely fixed -- there's still a bit of orange. Concerning.
- Wrote a short shell script to act like "top" for "perf". Used it to peer at V running Firefox. Should have done this years ago.
More of the same.
firstname.lastname@example.org last week
- Bug 1317173 - [non-e10s] Pop-ups opened in Private Browsing mode stay in history
- Bug 1317927 - Media caching needs to use origin attributes
- Bug 1319045 - Get rid of nsPrincipal::GetOriginFromURI
- Bug 1319370 - webkitdirectory - OS username or arbitrary local file disclosure
- Bug 1318727 - BroadcastChannel doesn't work with null principal
- Bug 1285898 - [e10s-multi] Localstorage "storage" event is not fired with multiple content processes
- Bug 1320275 - QuotaManager uses hal:GetCurrentBatteryInformation incorrectly
- Bug 1315905 - Cleanup Necko http security check
email@example.com last week
- Bug 1319437 - Ported Baseline (DOM) proxy stubs to CacheIR.
- Bug 1317703 - Ported Baseline native getter stub to CacheIR.
- Bug 1277368 - First JS::Result patch.
- Various other bugs: bug 1040402, bug 1320118.
Posted patches for:
- Bug 1320670, bug 1320145. This finishes porting Baseline's GETPROP stubs to CacheIR and also converts the GETELEM stubs to share the GETPROP code. The latter turned out to be a huge win in code size: it removes 1100+ lines of code and at the same time makes us optimize more cases. Very exciting this is working out as I hoped it would.
- Reviews, needinfos, travel to Hawaii
- Port the remaining GETELEM stubs to CacheIR.
- After that, we can implement a CacheIR compiler for the Ion GetProperty IC, so we can generate IC stubs from the same IR we currently feed into BaselineCacheIRCompiler. There are some challenges, but it should let us remove a lot of code that's currently duplicated between Ion and Baseline ICs.
firstname.lastname@example.org last week
- Started implementing cross origin wrapper
- Investigated switching certificate verification to webpki
- Learn how to codegen
- Update openssl dependencies (req. for updated cert verification)
- Go to Hawaii
email@example.com last week
- Internal goals reporting
- Kept hacking on Android builders
- Planning and management stuff
- TWIS & internal goals reporting
- Planning & management stuff
- srsly finish Android
firstname.lastname@example.org last week
- Audited Stylo reftest crashes, closed obsolete bugs, added root cause links.
- Started work on table cell
- Finish implementing
email@example.com last week
- Landed the fix to #633! We now share a lot more script threads: within a tab, similar-origin scripts will all be executed by the same script thread, so can share DOM objects. (https://github.com/servo/servo/pull/14211).
- Landed a PR to allow crash reporting without relying on there being a root pipeline. (https://github.com/servo/servo/pull/14173).
- Landed a PR to allow multiple -Z flags, which allows wpt testing to pass in -Z flags. (https://github.com/servo/servo/pull/14313).
- Landed the fix for a bug which was causing script threads to fail to spawn layout threads (https://github.com/servo/servo/pull/14344).
- Submitted a PR renaming constellation's ScriptChan to the more spec-compliant EventLoop (https://github.com/servo/servo/pull/14260).
- Submitted a PR for discarding inactive documents (https://github.com/servo/servo/pull/14312).
- Submitted a PR which allows us to share random-number generators, rather than having one per thread, which uses a file descriptor. (https://github.com/servo/servo/pull/14351).
- Thanksgiving: gave thanks for fixing #633.
- Attended Chicago TARDIS.
- Land PRs.
- Tidy up and document the constellation.
- Get ready for Hawaii.
firstname.lastname@example.org last week
- figured out a minimal test case for rr recording hangs
- tried to reproduce some intermittent failures under rr
- wrote processor for WPT structured log files
- prototyped initiating layout image loads from script
- reviewed bluetooth PRs
- reviewed script thread sharing PR
- reviewed about:blank PR
- dug into timeouts in about:blank PR
- updated blink patch to account for feedback from rbyers
- read source of macOS mach implementation to figure out ipc-channel issue
- tried to get TravisCI for osmesa working
- continue working on image loading prototype
- play with rr some more
email@example.com last week
- Finish passing the latest spec tests in bug 1316554 and close bug 1265461 (wasm spec testing) \o/.
- Fix an intermittent wasm error (shell only) in bug 1297901.
- Uplift overflow checking to aurora in bug 1316634.
- Found and fixed an Ion issue with phis marked as range guards in bug 1317675.
- Opened two indexeddb bugs related to wasm: bug 1319693 and bug 1319737
- Fuzz bug 1320226 for reporting two more OOMs.
- Bug 1320374: batch OMT wasm compilations.
- One day spent at a job fair in a local university, representing Mozilla.
- A bit of Cretonne's code reading and studying.
- Python training for 4 days; maybe making a wasm fuzzer as an example for the occasion.
- Fly out to the workweek.
firstname.lastname@example.org last week
email@example.com last week
- #14359 - Refactor drop-shadow filter
- Implementing URL filter
- #14332 - Added an "auto" keyword for flexbox's min-size
- Addressing review comments
- Rust beginner tutorial planning
- Study Rust book
- Implement URL filter
- Study how to implement drop-shadow in WR
- Flexbox min-size algorithm
- Rust trainings practice
- Rust internal training planning
firstname.lastname@example.org last week
- Found and fixed the WR patch which exposed a heap of Servo intermittents.
- Land WR update.
- Updated mac to latest OS to try and repro WR bugs, but failed.
- Many WR reviews.
- Prototype zoom support.
- Review and land several of the outstanding WR PRs.
- Finish up initial zoom support PR.
- Do some more profiling on various platforms and see why there are discrepancies.
email@example.com last week
firstname.lastname@example.org last week
HirVec<T>in hir::Expr. Speeds up compilation of some benchmarks by 1, 2 and 4%.
- Use a SmallVec for the TypeWalker stack. Speeds up compilation of some benchmarks by 1%.
- Avoid more unnecessary mk_ty calls in Ty::super_fold_with. Speeds up compilation of several rustc-benchmarks by 1--5%.
email@example.com last week
- Flattened an simplified the WebRender display list. Now it is much more similar to the Servo one.
- Reimplemented support for scroll to fragment in WebRender. Servo PR is pending.
- More cleanup of WebRender display list construction. Should help prevent API breaks in the future.
- Flatten the display list in the API https://github.com/servo/webrender/pull/552
- Remove establishes_3d_context field from StackingContex https://github.com/servo/webrender/pull/562
- Add PushScrollLayer and PopScrollLayer display list items https://github.com/servo/webrender/pull/567
- Add API for scrolling individual layers https://github.com/servo/webrender/pull/590
- Deal with changes to the WebRender API https://github.com/servo/servo/pull/14200
- Land scroll to fragment support
- Fix issue where borders and backgrounds of scrolling divs scroll instead of staying in place.
- Start working on making sure that scroll roots follow CSS containing block rules.
firstname.lastname@example.org 2 weeks ago
- reviewed (partially): bug 1305325 - Support missing keyframe whose computed offset is 0 or 1
- reviewed: bug 1316050 - Intermittent /web-animations/interfaces/AnimationEffectTiming/endDelay.html
- reviewed: bug 1317178 - Stylo: Use single_value_to_css in Servo_DeclarationBlock_SerializeOneValue
- reviewed: bug 1319336 - return early from some animation codepaths so that stylo tests make more progress
- wrote some thoughts: bug 1210796 - Display keyframes' timing-functions in the animation-inspector
- wrote some thoughts: bug 1309468 - CSS Animation's animation-timing-function curve should be visualized in summary graph
- filed: bug 1319989 - Don't show direction in tooltip when it is 'normal'
- investigation: bug 1318839 - 100% cpu usage if navigator.getGamepads() called
- Working on Web Animations timelines to make timelines work better with scroll-driven animations
- Lots of administrative work this week
- Investigated https://www.standu.ps/ since this service will be unavailable from next week!
- 1 day public holiday
email@example.com 2 weeks ago
- Bug 1316050 - Intermittent /web-animations/interfaces/AnimationEffectTiming/endDelay.html
- Bug 1319072 - Assertion failure: aCoeff >= 0.0 && aCoeff <= 1.0
- Bug 1318697 - Elements disappear after sliding in
- Bug 1305325 - Support missing keyframe whose computed offset is 0 or 1
- Bug 1273876 - Intermittent test_transitions_events.html
- Bug 1319384 - Setting a new animation whose from value does not equal to the base value triggers a new transition