[Merged by Bors] - Implement is_identifier_(start/part) using icu_properties

[jedel1043]

As mentioned in #2848 (comment), this uses our new default ICU4X data to replace char::is_start and char::is_continue from the boa_unicode crate with the icu_properties crate.

Note that this doesn't deprecate boa_unicode yet, since that'll require some discussion about how to proceed with a now unused sub-crate.

[github-actions]

Test262 conformance changes

Test result	main count	PR count	difference
Total	94,591	94,591	0
Passed	73,210	73,210	0
Ignored	17,530	17,530	0
Failed	3,851	3,851	0
Panics	0	0	0
Conformance	77.40%	77.40%	0.00%

[codecov]

Codecov Report

Merging #2865 (4835089) into main (739bd5a) will increase coverage by 0.03%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #2865      +/-   ##
==========================================
+ Coverage   50.92%   50.95%   +0.03%     
==========================================
  Files         419      419              
  Lines       41802    41867      +65     
==========================================
+ Hits        21288    21334      +46     
- Misses      20514    20533      +19     

Impacted Files	Coverage Δ
boa_parser/src/lexer/identifier.rs	96.55% <100.00%> (+0.63%)	⬆️

... and 13 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[Razican]

Looks good to me! Regarding boa_unicodde, we could maybe use cargo-tally (https://github.com/dtolnay/cargo-tally).

With this we could check if someone else uses it and deprecate our crate.

[jedel1043]

Looks good to me! Regarding boa_unicodde, we could maybe use cargo-tally (https://github.com/dtolnay/cargo-tally).

With this we could check if someone else uses it and deprecate our crate.

Apparently there are no dependents of boa_unicode apart from Boa and boa_engine (https://crates.io/crates/boa_unicode/reverse_dependencies), so I think it should be safe to deprecate it.

[Razican]

Looks good to me! Regarding boa_unicodde, we could maybe use cargo-tally (https://github.com/dtolnay/cargo-tally).
With this we could check if someone else uses it and deprecate our crate.

Apparently there are no dependents of boa_unicode apart from Boa and boa_engine (https://crates.io/crates/boa_unicode/reverse_dependencies), so I think it should be safe to deprecate it.

Let's do it! :)

[HalidOdat]

Looks good to me!

[jedel1043]

bors r+

[jedel1043]

Then, after this gets merged, we can open a new branch to bump icu_unicode to 0.16.1 with a description signaling its deprecated state, and delete the crate from our main branch.

[bors]

Pull request successfully merged into main.

Build succeeded:

• Build and Test (macos-latest)
• Build and Test (windows-latest)
• Coverage
• Misc

[raskad]

For the sake of documentation; I noticed that after this change our benchmarks for the parser looked better so I ran them and compared cebec9d to c330005.
This change seems to have improved the performance of the parser by a good amount:

Benchmark	Changes
Symbols (Parser)	[-13.850% -12.097% -10.548%]
For loop (Parser)	[-5.3454% -4.3440% -3.4117%]
Fibonacci (Parser)	[-13.609% -10.961% -8.5178%]
Object Creation (Parser)	[-8.1511% -6.9890% -5.7952%]
Static Object Property Access (Parser)	[-10.387% -9.2301% -8.0496%]
Dynamic Object Property Access (Parser)	[-7.9534% -6.8847% -5.7769%]
RegExp Literal Creation (Parser)	[-7.4906% -6.3700% -5.4132%]
RegExp Creation (Parser)	[-10.154% -9.1298% -7.9548%]
RegExp Literal (Parser)	[-7.7712% -6.7077% -5.5830%]
RegExp (Parser)	[-12.477% -11.538% -10.613%]
Array access (Parser)	[-15.038% -13.548% -11.900%]
Array creation (Parser)	[-17.589% -15.650% -13.799%]
Array pop (Parser)	[-8.6277% -7.0937% -5.5033%]
String concatenation (Parser)	[-12.413% -10.984% -9.6802%]
String comparison (Parser)	[-8.1121% -6.9242% -5.6435%]
String copy (Parser)	[-8.6093% -7.1169% -5.6620%]
Number Object Access (Parser)	[-10.555% -9.5270% -8.4492%]
Boolean Object Access (Parser)	[-17.107% -16.002% -14.850%]
String Object Access (Parser)	[-16.571% -15.397% -14.316%]
Arithmetic operations (Parser)	[+1.5678% +3.3078% +5.0827%]
Clean js (Parser)	[-7.3581% -6.1867% -4.6942%]
Mini js (Parser)	[-6.1803% -5.2065% -4.2830%]

cc @jedel1043

[jedel1043]

Nice! And it makes much sense, since icu_properties is hyperoptimized for char queries, whereas our old implementation was just a succession of queries using several tables and property matches that made the categorization work a lot slower.