(This newsletter is also available on our blog <https://mozilla-spidermonkey.github.io/blog/2020/03/12/newsletter-3.html>.) JavaScript đ New contributors
- Rob Rico tidied up <https://bugzilla.mozilla.org/show_bug.cgi?id=1605263> some code in the parser đ New features - Yulia implemented <https://bugzilla.mozilla.org/show_bug.cgi?id=1566143> the optional chaining (?.) operator (Firefox 74) - AndrĂ© implemented <https://bugzilla.mozilla.org/show_bug.cgi?id=1535804> public static class fields (Firefox 75) - AndrĂ© implemented <https://bugzilla.mozilla.org/show_bug.cgi?id=1613713> the Intl.Locale proposal (Firefox 75) đ SmooshMonkeyThe previous newsletter <https://mozilla-spidermonkey.github.io/blog/2020/01/10/newsletter-2.html#project-visage> introduced Visage, a new JavaScript frontend <https://github.com/mozilla-spidermonkey/jsparagus> weâre working on thatâs written in Rust. Visage has since been renamed to SmooshMonkey, a name thatâs known and well accepted <https://github.com/tc39/proposal-flatMap/pull/56> by the JavaScript community (#SmooshGate). After a dinner and discussions with project members, it got authoritatively renamed by speaking about it <https://docs.google.com/presentation/d/1pY5KQzFmayfZi5eLYI5673UVv9-lzdugbg-kOFpRRf0/edit#slide=id.g6dd8ea1b81_0_17> at the All-Hands. The team is making good progress <https://github.com/mozilla-spidermonkey/jsparagus/issues/304>: - SmooshMonkey has been integrated in Gecko <https://bugzilla.mozilla.org/show_bug.cgi?id=1612515> behind a configure flag and a runtime flag. - Passes 100% <https://github.com/mozilla-spidermonkey/jsparagus/milestone/1?closed=1> of SpiderMonkey tests (falling back on SpiderMonkeyâs current parser for non-implemented features). - Added stats about the project using Github CI. - The bytecode emitter has been improved to prevent generating bytecode which might have undefined behaviour. - Thereâs a new parser generator <https://github.com/mozilla-spidermonkey/jsparagus/pull/347> to support context-dependent aspects <https://github.com/mozilla-spidermonkey/jsparagus/issues/359> of the JavaScript grammar instead of exploding the number of states of the equivalent context-free grammar. âïž Stencil Progress on Project Stencil <https://bugzilla.mozilla.org/show_bug.cgi?id=1601332> is continuing. Huge thanks to AndrĂ© for helping knock three <https://bugzilla.mozilla.org/show_bug.cgi?id=1619007> blocking <https://bugzilla.mozilla.org/show_bug.cgi?id=1619010> bugs <https://bugzilla.mozilla.org/show_bug.cgi?id=1619008> off in quick succession! Matthew landed many patches to clean up <https://bugzilla.mozilla.org/show_bug.cgi?id=1611528> our compilation management data structures. He also added <https://bugzilla.mozilla.org/show_bug.cgi?id=1615728> a SourceExtent structure to store source information and changed the frontend to always defer <https://bugzilla.mozilla.org/show_bug.cgi?id=1610340> supported GC allocations. Kannan is working on <https://bugzilla.mozilla.org/show_bug.cgi?id=1592105> removing GC atom allocation from the frontend. The frontend uses atoms in many places and to make the frontend GC-free we need a different strategy for that. Caroline is working at <https://bugzilla.mozilla.org/show_bug.cgi?id=1620776> cleaning up the flags used throughout the frontend, to unify BaseScript::ImmutableFlags, CompileOptions, and FunctionBox flags into one representation. đ JSScript/LazyScript unification The JSScript/LazyScript unification <https://bugzilla.mozilla.org/show_bug.cgi?id=1529456> is nearing completion. Ted has landed patches for Firefox 75 to use the same GC TraceKind <https://bugzilla.mozilla.org/show_bug.cgi?id=1615143> for LazyScript and JSScript and after that was able to merge <https://bugzilla.mozilla.org/show_bug.cgi?id=1615145> the GC arenas. The is-lazy state has been moved <https://bugzilla.mozilla.org/show_bug.cgi?id=1591600> from JSFunction to BaseScript. The next big step is delazifying/relazifying scripts in place <https://bugzilla.mozilla.org/show_bug.cgi?id=1619803> so that we never have to keep both a JSScript and a LazyScript for a function. đž WarpBuilder Ion, our optimizing JIT, currently relies on a global Type Inference (TI) mechanism. Ion and TI have a number of shortcomings <https://bugzilla.mozilla.org/show_bug.cgi?id=1613592#c0> so weâre experimenting with a much simpler MIR builder for Ion thatâs based on Baseline ICs (CacheIR) instead of TI. If this works out it will let us delete some of our most complicated code, allow us to do more work off-thread, and result in memory usage reductions and performance improvements across the engine. The past weeks Jan landed patches preparing <https://bugzilla.mozilla.org/show_bug.cgi?id=1613594> for this <https://bugzilla.mozilla.org/show_bug.cgi?id=1616188> and added <https://bugzilla.mozilla.org/show_bug.cgi?id=1617564> a very primitive WarpBuilder implementation thatâs able to build MIR off-thread. Heâs now adding support <https://bugzilla.mozilla.org/show_bug.cgi?id=1618198> for more bytecode instructions. â© Regular expression engine update Iain is working on upstreaming <https://chromium-review.googlesource.com/c/v8/v8/+/2072858> some changes to v8 to improve case-insensitive match support. He also started posting patches to implement <https://bugzilla.mozilla.org/show_bug.cgi?id=1620020> shim definitions to make the code work in SpiderMonkey. đ JIT optimizations for classes and spread calls AndrĂ© added Ion support <https://bugzilla.mozilla.org/show_bug.cgi?id=1378189> for derived class constructors. He then made it possible <https://bugzilla.mozilla.org/show_bug.cgi?id=1557765> for Ion to inline class constructors. He also added Ion support <https://bugzilla.mozilla.org/show_bug.cgi?id=1619343> for spread-new and spread-super calls. âïž Miscellaneous - Steve started converting <https://bugzilla.mozilla.org/show_bug.cgi?id=1619475> our JS test suites to support Python 3. - Steve also added <https://bugzilla.mozilla.org/show_bug.cgi?id=1614518> a mach hazards command to make it easier to run the GC rooting analysis locally. - Jon enabled (browser) tests <https://bugzilla.mozilla.org/show_bug.cgi?id=1616230> for JS WeakRefs <https://github.com/tc39/proposal-weakrefs>. - AndrĂ© added <https://bugzilla.mozilla.org/show_bug.cgi?id=1530372> nursery allocation support for BigInts. - Iain noticed the C++ interpreter was not using computed goto <https://gcc.gnu.org/onlinedocs/gcc/Labels-as-Values.html> on Windows (supported since the switch to clang-cl) and fixed this <https://bugzilla.mozilla.org/show_bug.cgi?id=1609229>. - IRC has been shut down. You can now find us in chat.mozilla.orgâs Spidermonkey room <https://chat.mozilla.org/#/room/#spidermonkey:mozilla.org>. WebAssembly - Ryan added support for using Cranelift when reference types are enabled <https://bugzilla.mozilla.org/show_bug.cgi?id=1574865>. Reference types allow WebAssembly to pass JavaScript values as function arguments and return values, to store them in local variables, and to load and store them in tables <https://webassembly.github.io/reference-types/core/syntax/instructions.html#syntax-instr-table>. Cranelift can now emit annotations that inform the JavaScript garbage collector of value locations, so that the GC can trace and relocate object pointers. - Andy from Igalia added support for functions that return multiple values <https://github.com/WebAssembly/multi-value/blob/master/proposals/multi-value/Overview.md> to the WebAssembly baseline compiler <https://bugzilla.mozilla.org/show_bug.cgi?id=1603140>; when the corresponding implementation lands in the optimizing compiler <https://bugzilla.mozilla.org/show_bug.cgi?id=1607200> (Firefox 76), the feature will finally be able to ride the train out of Nightly. - Work is ongoing to replace Craneliftâs instruction selection mechanism as well as register allocation. We can now compile basic wasm programs with either Wasmtime or Spidermonkey using Craneliftâs new pipeline. - We also bid a fond farewell to the #wasm IRC channel; you can now find the SpiderMonkey WebAssembly team over on chat.mozilla.orgâs WebAssembly room <https://chat.mozilla.org/#/room/#webassembly:mozilla.org>. _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
