[Webkit-unassigned] [Bug 6257] Throw errors on invalid expressions/support non-PCRE regexp (KJS merge)

[bugzilla-daemon at opendarwin.org]

http://bugzilla.opendarwin.org/show_bug.cgi?id=6257





------- Additional Comments From ggaren at apple.com  2006-01-11 09:50 -------
RegExp("\\u" + "0020") is a really interesting case -- something I hadn't considered. Both Firefox and 
MacIE honor it, and a quick reading of the spec suggests it's required, so, yes, that rules out the lexer 
solution. 

I'm not sure if we still care about supporting the POSIX library. Darin? Are there any platforms on which 
we can't use PCRE? Regardless, I think the 2-pass approach is worth it. It would just require factoring 
the Unicode escape converting code into its own function and calling that function as a fallback when 
the regular expression compiler returns an error.

With your patch applied, ecma_3/RegExp/regress-119909.js doesn't fail because it crashes. Rather, 
where we used to silently ignore the regular expression, we now throw an error. And our test suite 
interprets thrown errors as failures. Since we expect this test to throw an error, the simplest solution is 
to surround the test code in a try/catch block. The expectation for the test can remain the same.

-- 
Configure bugmail: http://bugzilla.opendarwin.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.