Demangle symbols when reading them out of the symbol table.

[mstange]

Doing demangling in perf.html has the following advantages:

• If we add code to Firefox that dumps symbol tables from binaries (which I'm currently doing for Firefox on Android in order to get symbols for Android system libraries), Firefox doesn't have to include a demangler. This saves on code size.
• Mangled names are usually a bit shorter than demangled names. The symbol tables we store will be a bit more compact if we don't require the strings in them to be in a demangled form.
• Demangling takes a bit of time. Only demangling the strings that we actually need to symbolicate a given profile saves time by not demangling everything else.

This also adds our first wasm module and updates the configuration so
that that's possible.
I've used wasm-pack to publish a gecko-profiler-demangle module to npm
instead of copying the generated code into this repository. I'm not sure
if that was a good or a bad idea; I mostly did it so that our Flow and
ESLint rules wouldn't be applied to the wasm-bindgen generated code.

I've also needed to autogenerate a flow libdef for this module.
wasm-bindgen doesn't create those automatically yet:
rustwasm/wasm-bindgen#180

I could have created one manually and included it in the published
package, but I don't know if wasm-pack would have included my additional
files in the published package (via wasm-pack publish).

---

Todo:

• get tests running again - they fail because of the async import(), Our test infrastructure can't deal with async import() #1204
• [ ] add tests that verify that things get demangled (impossible because we don't have the demangler in the test suite. We could mock some fake demangling and test that it happens when getting symbols out of symbol tables, but it would be silly.)

[mstange]

This breaks the production build due to NekR/offline-plugin#408.

[codecov]

Codecov Report

Merging #1203 into master will increase coverage by <.01%.
The diff coverage is 75%.

@@            Coverage Diff             @@
##           master    #1203      +/-   ##
==========================================
+ Coverage   77.82%   77.82%   +<.01%     
==========================================
  Files         151      151              
  Lines       10023    10025       +2     
  Branches     2449     2449              
==========================================
+ Hits         7800     7802       +2     
  Misses       2003     2003              
  Partials      220      220

Impacted Files	Coverage Δ
src/index.js	0% <0%> (ø)	⬆️
src/profile-logic/symbol-store.js	93.82% <100%> (+0.15%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5498749...c4c915c. Read the comment docs.

[mstange]

IT'S GREEEN!!!!

[gregtatum]

Thanks, this works for me! I left some comments that I feel should be addressed before landing, but I don't think they need a re-review.

I also looked at the gecko-profiler-demangle code which looked decently straightforward for the needs here.

[gregtatum]

Is there a bug somewhere to track this feature? If so it would be nice to link it here.

[mstange]

Added:

// The long-term path to make this work correctly is to wait for node to
// support ES6 modules (and WASM as ES6 modules) natively [1]. It's possible
// that in the medium term, wasm-bindgen will get support for outputting JS
// files which work in both webpack and in node natively [2].
// [1] https://medium.com/@giltayar/native-es-modules-in-nodejs-status-and-future-directions-part-i-ee5ea3001f71
// [2] https://github.com/rustwasm/wasm-bindgen/issues/233