v0.31.3

Fixes

• Properly name more shim functions
• Fix an IE bug where the layout engine internally calls the userland Object.getOwnPropertyNames (#333)
• Ensure Map.prototype[Symbol.iterator] === Map.prototype.entries
• Ensure Set.prototype[Symbol.iterator] === Set.prototype.values

Security/Robustness

• - Cache more native methods in case they're overwritten later

Performance

• Add more Map/Set fast paths for more primitives: boolean, null, undefined

Tests

• Object.assign pending exceptions: IE 9 preventExtensions doesn't throw, even in strict mode
• IE 11 has native Map/Set, but it takes an optional function, not an optional iterable, in the constructor
• Add more "exists" early bailouts, to declutter native test results
• Test up to io.js v2.2

Docs

• Alphabetize shim lists in the README

Dependencies

• Update mocha, es5-shim, uglify-js, jshint

Refactoring

• Style cleanups

v0.31.2

Bugfix

• Ensure that wrapped ES5 Array.prototype methods return the proper value. (#341)

v0.31.1

Bugfix

• RegExp should work properly as a wrapper (#340)

v0.31.0

New shims:

• All Array.prototype methods should use ToLength, not ToUint32, on this.length.
• Preserve and use original ES5 Array.prototype functions (for later shimming)

Tests:

• Make String#{startsWith, endsWith, includes} tests a bit more granular.
• Fix Map/Set invalid receiver error messages for WebKit

Dependencies:

• Update grunt-saucelabs, jscs

v0.30.0

Breaking change

• Map and Set prototype methods are not generic, and must only be called on valid Map and Set objects.

Performance

• Use the native Number#clz (in Safari 8, eg) inside Math.clz32

v0.29.0

Why the minor version was bumped

• Ensure that ES3 browsers get both Object.is and Object.assign
• Ensuring Number.parseInt === parseInt (failed in FF 37)

Enhancements

• Improve Object.assign to avoid leaking arguments in v8
• a little more accurate Math.cbrt (#335)
• Avoid reassigning argument variables to avoid deoptimizations
• Ensure that shallowly wrapped Maps’ and Sets’ prototypes aren't one level too far away.

Testing

• Test on io.js v1.7 and v1.8
• Update chai and use new matchers
• Test cleanups
• Adding Symbol.unscopables tests
• Adding tests to ensure that default iterators on builtins === the appropriate prototype function.

v0.28.2

• Map and Set should have an arity of 0.

v0.28.1

• Ensure Object.assign only includes enumerable Symbols. (dc23efa#commitcomment-10684061)

v0.28.0

• Ensure Object.assign also includes Symbols.
• Make sure to clobber Firefox 37's very slow native Object.assign, that has "pending exception" logic.
• Adding much more granular Set/Map acceptance tests and replacements, to preserve as much of the original implementation as possible. (#326, #328)
  • Lots of test additions and cleanup
  • Fill in (and fix) missing name, arity, and enumerability tests.
  • Using property matcher for a more helpful failure message.
  • Make sure this test doesn't fail if Array#values doesn't exist yet.
  • Make this @@iterator test not depend on Array#values, and properly skip tests if the symbol isn't available.
• Update Math.fround with a much smaller implementation (#332)
• Lock uglify-js down to v2.4.17, since v2.4.18 and v2.4.19 have a breaking change.
• Update es5-shim, mocha, grunt-contrib-connect, chai, jshint
• IE 11 TP has a broken String.raw implementation
• Overwriting some imprecise Math functions on IE 11 TP.
• Overwrite Math.imul in Safari 8 to report the correct length.
• Fix Math.round for very large numbers
• Don't rely on shims in tests, for better native failure checking.
• Shim Object.is in ES3 environments, and add tests.
• Test the native Object.assign prior to shimming it.
• Tweak the travis-ci config to make a separate "lint only" test run.
• Fix Firefox 4 test failures: ensure RegExp global aliases starting with "$" exist.
• more efficient Math.clz32 (#327)
• Fix Webkit nightly bugs with Array.from and Array.of.
• Make sure shims that depend on Number.isNaN and Number.isFinite will always work.
• The latest Webkit nightly has a bug with String#includes and a position arg of Infinity.
• Webkit r181855 has a noncompliant String#startsWith and String#endsWith
• Clean up README; add more accurate note about es5-shim.
• Updating the String.raw code to be more in line with the changes in RC2/Rev 35 of the spec.

v0.27.1

• Revert Array#slice changes. (#322)
• Test on io.js v1.4