Implementation of strict mode in SetElement.

Implementation of strict mode in SetElement.

Passing strict mode throughout SetElement.
Throw if assigning to read only element.
Adding tests for element assignment in strict mode.
Fix tests for strict mode SetElement.

BUG=http://code.google.com/p/v8/issues/detail?id=1220
TEST=test/mjsunit/strict-mode.js

[Martin Maly, 2011-03-03 06:07:56]

The remainder of the 'assignment to read only property" implementation -
SetElement.

Thank you!
Martin

http://codereview.chromium.org/6613005/diff/2001/src/debug.cc
File src/debug.cc (right):

http://codereview.chromium.org/6613005/diff/2001/src/debug.cc#newcode1019
src/debug.cc:1019: kNonStrictMode);
Debugger code, not using strict mode.

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc#newcode8675
src/runtime.cc:8675: SetElement(jsobject, index1, tmp2, kNonStrictMode);
Runtime_SwapElements - used by quick sort. I couldn't find any mention that the
array sorting should behave strictly so I am keeping it non-strict.

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc#newcode11502
src/runtime.cc:11502: SetElement(result, index++, pair, kNonStrictMode);        
              \
Debug code only, no need for strict mode.

http://codereview.chromium.org/6613005/diff/2002/src/handles.cc
File src/handles.cc (right):

http://codereview.chromium.org/6613005/diff/2002/src/handles.cc#newcode269
src/handles.cc:269: object, key, value, attributes),
fixed the whitespace :)

[Lasse Reichstein, 2011-03-03 09:39:50]

LGTM

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc#newcode8675
src/runtime.cc:8675: SetElement(jsobject, index1, tmp2, kNonStrictMode);
It's used only internally in the Array.prototype.sort implementation, and that's
defined as doing a sequence of [[Get]], [[Put]], [[Delete]] and calls to the
compare function, where the "throw" argument to [[Put]] must be true.
I.e., we should throw if [[Put]](i,v,true) would throw (ES5 8.12.5, refers to
[[CanPut]], 8.12.14) which is when there is either no setter for an accessor
property or if [[Writable]] is false for a data accessor.
Or if the object isn't extensible (frozen), unless it inherits a setter
accessor.
Or some more twisted logic.

(I think the [[CanPut]] spec is wrong, it sees to only consider the immediate
prototype in the prototype chain).

In any case, I would go kStrictMode here and see if it breaks anything.
Never sort with write-only properties in your array!

[Lasse Reichstein, 2011-03-03 09:44:53]

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc#newcode8675
src/runtime.cc:8675: SetElement(jsobject, index1, tmp2, kNonStrictMode);
My bad, the spec is fine. It uses [[GetProperty]] on the prototype object, not
[[GetOwnProperty]], so it does handle the remainder of the prototype chain.

[Martin Maly, 2011-03-04 00:36:09]

Thanks for the feedback. Landed. Hopefully webkit tests will pass (ran the set
that triggered last locally with success).

Thanks!
Martin

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/6613005/diff/2001/src/runtime.cc#newcode8675
src/runtime.cc:8675: SetElement(jsobject, index1, tmp2, kNonStrictMode);
On 2011/03/03 09:39:50, Lasse Reichstein wrote:
> It's used only internally in the Array.prototype.sort implementation, and
that's
> defined as doing a sequence of [[Get]], [[Put]], [[Delete]] and calls to the
> compare function, where the "throw" argument to [[Put]] must be true.
> I.e., we should throw if [[Put]](i,v,true) would throw (ES5 8.12.5, refers to
> [[CanPut]], 8.12.14) which is when there is either no setter for an accessor
> property or if [[Writable]] is false for a data accessor.
> Or if the object isn't extensible (frozen), unless it inherits a setter
> accessor.
> Or some more twisted logic.
> 
> (I think the [[CanPut]] spec is wrong, it sees to only consider the immediate
> prototype in the prototype chain).
> 
> In any case, I would go kStrictMode here and see if it breaks anything.
> Never sort with write-only properties in your array!

Done. Strict mode worked just fine.