[webkit-reviews] review requested: [Bug 9179] Implement select.options.add() method : [Attachment 9325] Patch v3

[bugzilla-request-daemon at opendarwin.org]

David Kilzer (ddkilzer) <ddkilzer at kilzer.net> has asked  for review:
Bug 9179: Implement select.options.add() method
http://bugzilla.opendarwin.org/show_bug.cgi?id=9179

Attachment 9325: Patch v3
http://bugzilla.opendarwin.org/attachment.cgi?id=9325&action=edit

------- Additional Comments from David Kilzer (ddkilzer) <ddkilzer at kilzer.net>
Changes since Patch v2:

- Added more test cases specifically targeting -Infinity, NaN and +Infinity in
fast/dom/select-selectedIndex-multiple.html and
fast/js/select-options-add.html.
- Made sure fast/dom/select-selectedIndex.html was kept in sync with test cases
from select-selectedIndex-multiple.html.
- Overloaded KJS::JSValue::toInt32() to take a boolean 'Ok' argument to flag
when zero is returned for a float with a value of +-Infinity or NaN.
- Setting the options.length to a negative value now throws an exception. 
(This mimics MSIE 6, but differs from Firefox 1.5.0.x.)
- Passing anything other than an option element to options.add() causes an
exception to be thrown for all cases now (including +-Infinity and NaN, which
were being valued to zero before; this is where the new toInt32() method is
used to trap these values as errors).
- Changed overloaded function feature in CodeGeneratorJS.pm to use "argCount <"
instead of "argCount ==" comparison.
- Changed CanFailTypeConversion() in CodeGeneratorJS.pm to return true for
"unsigned long" and "long" types.  (They now use the new toInt32() overloaded
method.)
- Changed JSValueToNative() in CodeGeneratorJS.pm to use new toInt32()
overloaded method.

All layout tests still pass with these changes.  Will upload Firefox 1.5.0.5rc3
and MSIE 6 test results next.