Skip to content

Latest commit

 

History

History
268 lines (217 loc) · 14.1 KB

CHANGELOG.md

File metadata and controls

268 lines (217 loc) · 14.1 KB

CHANGELOG

v0.7.0 brings a REPL, Improved parser messages and a new parser! This is now the default behaviour of Boa, so running Boa without a file argument will bring you into a javascript shell.
Tests have also been moved to their own files, we had a lot of tests in some modules so it was time to separate.

New Parser

Most of the work in this release has been on rewriting the parser. A big task taken on by HalidOdat, Razican and myself.

The majority of the old parser was 1 big function (called parse) which had some pattern matching on each token coming in.
The easy branches could generate expressions (which were basically AST Nodes), the more involved branches would recursively call into the same function, until eventually you had an expression generated.

This only worked so far, eventually debugging parsing problems were difficult, also more bugs were being raised against the parser which couldn't be fixed.

We decided to break the parser into more of a state-machine. The initial decision for this was inspired by Fedor Indutny who did a talk at (the last) JSConf EU about how he broke up the old node-parser to make it more maintanable. He goes into more detail here https://www.youtube.com/watch?v=x3k_5Mi66sY&feature=youtu.be&t=530

The new parser has functions to match the states of parsing in the spec. For example https://tc39.es/ecma262/#prod-VariableDeclaration has a matching function read_variable_declaration. This not only makes it better to maintain but easier for new contributors to get involed, as following the parsing logic of the spec is easier than before.

Once finished some optimisations were added by HalidOdat to use references to the tokens instead of cloning them each time we take them from the lexer.
This works because the tokens live just as long as the parser operations do, so we don't need to copy the tokens.
What this brings is a huge performance boost, the parser is 67% faster than before!

Parser Improvement

Feature enhancements:

  • FEATURE #281: Rebuild the parser (@jasonwilliams, @Razican, @HalidOdat)
  • FEATURE #278: Added the ability to dump the token stream or ast in bin. (@HalidOdat)
  • FEATURE #253: Implement Array.isArray (@cisen)
  • FEATURE Switch to normal output instead of debugged output (stdout/stdout) (@jasonwilliams)
  • FEATURE #258: Moved test modules to their own files (@Razican)
  • FEATURE #267: Add print & REPL functionality to CLI (@JohnDoneth)
  • FEATURE #268: Addition of forEach() (@jasonwilliams) (@xSke)
  • FEATURE #262: Implement Array.prototype.filter (@Nickforall)
  • FEATURE #261: Improved parser error messages (@Razican)
  • FEATURE #277: Add a logo to the project (@HalidOdat)
  • FEATURE #260: Add methods with f64 std equivelant to Math object (@Nickforall)

Bug fixes:

  • BUG #249: fix(parser): handle trailing comma in object literals (@gomesalexandre)
  • BUG #244: Fixed more Lexer Panics (@adumbidiot)
  • BUG #256: Fixed comments lexing (@Razican)
  • BUG #251: Fixed empty returns (@Razican)
  • BUG #272: Fix parsing of floats that start with a zero (@Nickforall)
  • BUG #240: Fix parser panic
  • BUG #273: new Class().method() has incorrect precedence

Documentation Updates:

  • DOC #297: Better user contributed documentation

The lexer has had several fixes in this release, including how it parses numbers, scientific notation should be improved.
On top of that the lexer no longer panics on errors including Syntax Errors (thanks @adumbidiot), instead you get some output on where the error happened.

Moving to a workspace architecture

Boa offers both a CLI and a library, initially these were all in the same binary. The downside is those who want to embed boa as-is end up with all of the command-line dependencies.
So the time has come to separate out the two, this is normal procedure, this should be analogous to ripgrep and the regex crate.
Cargo has great support for workspaces, so this shouldn't be an issue.

Benchmarks

We now have benchmarks which run against master!
Thanks to Github Actions these will run automatically a commit is merged.

Feature enhancements:

Bug fixes:

  • BUG #222: Fixed clippy errors (@IovoslavIovchev)
  • BUG #228: [lexer: single-line-comment] Fix bug when single line comment is last line of file (@croraf)
  • BUG #229: Replace error throwing with panic in "Lexer::next()" (@croraf)
  • BUG #232/BUG #238: Clippy checking has been scaled right back to just Perf and Style (@jasonwilliams)
  • BUG #227: Array.prototype.toString should be called by ES value (@cisen)
  • BUG #242: Fixed some panics in the lexer (@adumbidiot)
  • BUG #235: Fixed arithmetic operations with no space (@gomesalexandre)
  • BUG #245: Fixed parsing of floats with scientific notation (@adumbidiot)

Feature enhancements:

  • FEATURE #151: Implement the Rest/Spread operator (functions and arrays).
  • FEATURE #193: Implement macro for setting builtin functions
  • FEATURE #211: Better Display support for all Objects (pretty printing)

Feature enhancements:

  • FEATURE #119: Introduce realm struct to hold realm context and global object.
  • FEATURE #89: Implement exponentiation operator. Thanks @arbroween
  • FEATURE #47: Add tests for comments in source code. Thanks @Emanon42
  • FEATURE #137: Use Monaco theme for the demo page
  • FEATURE #114: String.match(regExp) is implemented (@muskuloes)
  • FEATURE #115: String.matchAll(regExp) is implemented (@bojan88)
  • FEATURE #163: Implement Array.prototype.every() (@letmutx)
  • FEATURE #165: Implement Array.prototype.find() (@letmutx)
  • FEATURE #166: Implement Array.prototype.findIndex() (@felipe-fg)
  • FEATURE #39: Implement block scoped variable declarations (@barskern)
  • FEATURE #161: Enable obj[key] = value syntax.
  • FEATURE #179: Implement the Tilde operator (@letmutx)
  • FEATURE #189: Implement Array.prototype.includes (incl tests) (@simonbrahan)
  • FEATURE #180: Implement Array.prototype.slice (@muskuloes @letmutx)
  • FEATURE #152: Short Function syntax (no arguments)
  • FEATURE #164: Implement Array.prototype.fill() (@bojan88)
  • Array tests: Tests implemented for shift, unshift and reverse, pop and push (@muskuloes)
  • Demo page has been improved, new font plus change on input. Thanks @WofWca
  • FEATURE #182: Implement some Number prototype methods (incl tests) (@pop)
  • FEATURE #34: Number object and Constructore are implemented (including methods) (@pop)
  • FEATURE #194: Array.prototype.map (@IovoslavIovchev)
  • FEATURE #90: Symbol Implementation (@jasonwilliams)

Bug fixes:

  • BUG #113: Unassigned variables have default of undefined (@pop)
  • BUG #61: Clippy warnings/errors fixed (@korpen)
  • BUG #147: Updated object global
  • BUG #154: Correctly handle all whitespaces within the lexer
  • Tidy up Globals being added to Global Object. Thanks @DomParfitt

0.4.0 (2019-09-25)

v0.4.0 brings quite a big release. The biggest feature to land is the support of regular expressions.
Functions now have the arguments object supported and we have a debugging section in the docs.

Feature enhancements:

  • FEATURE #6: Support for regex literals. (Big thanks @999eagle)
  • FEATURE #13: toLowerCase, toUpperCase, substring, substr and valueOf implemented (thanks @arbroween)
  • Support for arguments object within functions
  • StringData instead of PrimitieData to match spec
  • Native function signatures changed, operations added to match spec
  • Primitives can now be boxed/unboxed when methods are ran on them
  • Spelling edits (thanks @someguynamedmatt)
  • Ability to set global values before interpreter starts (thanks @999eagle)
  • Assign operators implemented (thanks @oll3)

Bug fixes:

  • BUG #57: Fixed issue with stackoverflow by implementing early returns.
  • Allow to re-assign value to an existing binding. (Thanks @oll3)

0.3.0 (2019-07-26)

  • UnexpectedKeyword(Else) bug fixed jasonwilliams#38
  • Contributing guide added
  • Ability to specify file - Thanks @callumquick
  • Travis fixes
  • Parser Tests - Thanks @Razican
  • Migrate to dyn traits - Thanks @Atul9
  • Added implementations for Array.prototype: concat(), push(), pop() and join() - Thanks @callumquick
  • Some clippy Issues fixed - Thanks @Razican
  • Objects have been refactored to use structs which are more closely aligned with the specification
  • Benchmarks have been added
  • String and Array specific console.log formats - Thanks @callumquick
  • isPropertyKey implementation added - Thanks @KrisChambers
  • Unit Tests for Array and Strings - Thanks @GalAster
  • typo fix - Thanks @palerdot
  • dist cleanup, thanks @zgotsch

0.2.1 (2019-06-30)

Some String prototype methods are implemented.
Thanks to @lennartbuit we have trim/trimStart/trimEnd added to the string prototype

Feature enhancements:

Bug fixes:

  • Plenty

0.2.0 (2019-06-10)

Working state reached

  • Tests on the lexer, conforms with puncturators and keywords from TC39 specification
  • wasm-bindgen added with working demo in Web Assembly
  • snapshot of boa in a working state for the first time