This change adds 'var' declarations for global variables to allow the tests to run in strict mode (see issue #35).
Extra care was taken to ensure the changes do not alter the test behavior, for example when implicit creation of global variables are part of the test.
Note: The change does not fix all strict mode errors due to missing 'var' declarations.
The with statement tests were novel, but the tests for
property accessors throwing errors for undefined & null
were duplicates with S11.2.1_A3_T4/5 respectively.
Sbp_A1_T1 to Sbp_A5_T2, Sbp_12, Sbp_7*: Functions are now valid in blocks, tests syntactically invalid anyway
Sbp_A10_T1 to Sbp_A10_T2: Tests are broken
add test to repro #114
fix issue #114
factor out PercentFormat function
new fn write; start factoring summary output
fix tests to match
consolidate "Failed Tests" message
add test for successful run
consolidate all_succeeded message
consolidate summary line
consolidate ntests
failing negative tests
squeeze whitespace
add more tests of Promise.all
additional Promise test
add tests of Promise.prototype
add some tests for Promise.race
additional Promise tests
add Promise.reject and Promise.resolve tests
correct test description
rename badly-named files
use current license and minor style cleanup
correct understanding of undefined as thisArgument
incorporate line notes
Is this enough to make a constructor in ES6?
more accurate use of resolved,fulfilled etc.
remove some redundant tests
add new tests
remove "constant array" unclear language
better description
Update S25.4.2.1_A3.2_T1.js
address dangling ()
changes per @anba line notes
- if GetIterator is abrupt, the Promise.race / Promise.all should reject
- if Promise.race is called with nonconforming constructor as 'this',
TypeError should be thrown (cannot reject if exeption is thrown from
NewPromiseCapability; no promise exists yet...)
correct description of "this" testing in callbacks
fix whitespace, formatting
remove tab
add "next-turn" checking to sequencers
parseTestRecord.py:
- recover from ImportError when YAML not defined
- use monkeyYaml as backup
monkeyYaml.py:
- parser for subset of YAML used in test262 frontmatter
test_monkeyYaml.py:
- unit tests
CONTRIBUTING.md
- document `timeout` tag
- reorder tags in frontmatter doc
- minor cleanups
- minor fixes
- add style note
- reformat flags
- remove discussion of obsolete $INCLUDE
- incorporate line notes from @domenic
- integrate additional comments
- add links back, move arg notes down
- Raise outline level by one
README.md
- link to CONTRIBUTING
Test of spec language from 15.4.4.4,
5.b.iii.3.b states 'Call the [[DefineOwnProeprty]] internal method of A'
which means that the result of Array.prototype.concat
should return `true` from `hasOwnProperty` for every property
created by `concat`, up to `length`
Set length to 3.
a[0] is set to 0 on a
a[1] is unset
a[2] is set to 2 via a prototype of a
add assertions for values of b[0], b[1], b[2]
add assertions for b.hasOwnProperty('2')
Check with Array.prototype and Object.prototype, separate tests
fix typo in assertion message
test was correct, but assertion message was incorrect
1. use negative regex (instead of ".") to check iframeError
2. make Test262Error.prototype.toString() always include
the string Test262Error (no spaces) so it matches /Test262Error/
3. modify check for supportsWindowOnerror - require that
first argument (message) to onerror() contains the error name
by checking for /Error:/.
Change (3) above forces IE11 onto the !supportsWindowOnerror path.
Test262 tests are run inside an eval() on IE11, and errors are
caught and explicitly have toString() called. Without this,
IE11 passes only the `message` property to onerror(), and regexes
that expect to match error name fail.
sth: revert to simple `onerror` checking
gs: refactor function `$DONE`
refactor logic tree
add support for async test failures
ed.js: crude support for error typing
S8.7.1_A2: cache result of delete
`delete` has a side-effect and its
return value depends on this; cache result of
delete so when reporting result, we are always
reporting the result that made us fail, not the
result of a new delete
[pedantic]
ed: explicitly pass cooked error to $DONE
gs: only let $DONE be called once