- a lot of JS harness code written in strings have been moved out to actual physical files
such as ed.js (syntax error detection for globally scoped tests) and gs.js (global scope test
case validator). This change makes it far easier to maintain the test harness code
- reorganized helper.js providing a clear indication which methods are used by external objects,
which are implementation details, and which are unequivocally test262-specific. I've also added,
openErrorWindow, which will be used to open a descriptive error message window for each test case
failure reported on the 'Run' tab
- improved the error message for syntax errors occurring when a test case fails to load
- sta.js no longer tries to pickle all helper functions it contains! Instead, we load the file
directly from sth.js. The performance of fnGlobalObject has been improved. Finally, the ES5Harness
object has been moved from sth.js (in a string) to here
- sth.js now has a browser implementer hook, controller.implementerHook, which allows browser implementers
to handle test case failures in their own way (e.g., log to the filesystem). The 'run' function was
basically re-written
Added 37 new test cases from the "IE Test Center" Build release. There were 14 modifications to
existing test cases as well. Refactored SputnikGlobalScope.js such that test case paths are now used
as indices into the GlobalScopeTests array.
TestCasePackager.py had the concept of templated test harnesses introduced - see templates\runner.test262.html.
Also added support for one HTML test harness per ES5 chapter. Last but not least, TestCasePackagerConfig.py
now has a 'source control' abstraction class which abstracts away source control adds|edits when dynamically
generating *.json and *.html test chapters.
tests covering a given chapter, N, without delving into subsections. This in turn
broken the Results page which was by default displaying individual results for chapters
13(.0) and 14(.0). Fixed.
directory next*
SputnikConverter:
- ES5TestCase.cs
* Case of JSON-based property names was wrong. Fixed
* Use the tests' 'id' instead of 'path' as the GlobalScopeTests keys
* Added 'id' and 'path' as Global Scope test metadata. The correct/clean solution here is to simply use 'path'
as the key to GlobalScopeTests, but this refactoring needs to be undertaken later when we can convert the
'normal' test cases over to use 'path' as the key as well
* Turns out we cannot depend on the message received by window.onerror to have some form of "syntax"
contained within it. Instead, we'll just rely on the regular expression ".", matches any one character,
for the short term
- SputnikTestCase.cs
* Same case issue as for ES5TestCases.cs
* Don't trust the Sputnik metadata for the ES5 section name or even test case id to be correct. Instead,
generate this information from the file path of the test case
TestCasePackager.py:
- added a new global, GLOBAL_SCOPE_FILES, which is a list of JS files found directly under test\suite\*.js
which include metadata for so-called globally scoped tests. These files are imported directly by the HTML
test harness
test\suite\*:
- regenerated Sputnik tests based on new converter
default.html:
- import SputnikGlobalScope.js. Really TestCasePackager.py should generate the global scope imports to
default.html automatically...
website\resources\scripts\testcases\*:
- test cases have shuffled from existing *.json files into globalscope.json
their test case objects. Created an automated tool, FixPathsAndIds.py,
to fix these.
TestCasePackager.py no longer throws when it encounters a non-*.js
file where it expects test cases. Instead, it simply ignores it.
rationale behind the addition of this file is it's meant to replace
TestCasePackager.ps1 outright and will be runnable on all major
operating systems. There's still a bit of cleanup work needed as
the Python port was hardly optimized for the language.
A few side effects of this change:
- it now takes under ten seconds to generate the test262 website (PS
version was taking over a minute)
- generated *.json files are ~5% smaller in size (PS was forcing
'\r\n' and the Python version preserves the tests' newlines whether
they're '\r\n' or '\n')
ES5 chapter. That is, ecma-262-toc.xml now has '.0' sections for all chapters
and I've manually modified (Sputnik Ch. 13 & 14) test case id's and paths
to utilize the '.0' suffix. Long term, SputnikConverter needs to perform
this transformation itself though.
David Fugate
Mark Miller