Add runtime function for string to array conversion.

Add runtime function for string to array conversion.

Committed: http://code.google.com/p/v8/source/detail?r=4022

[antonm, 2010-03-04 13:23:15]

drive-by comment

http://codereview.chromium.org/669060/diff/1/3
File src/factory.cc (right):

http://codereview.chromium.org/669060/diff/1/3#newcode47
src/factory.cc:47: ASSERT(0 <= size);
I am not sure we should ever allocate empty fixed uninitialized arrays.

[Vitaly Repeshko, 2010-03-04 13:25:58]

http://codereview.chromium.org/669060/diff/1/3
File src/factory.cc (right):

http://codereview.chromium.org/669060/diff/1/3#newcode47
src/factory.cc:47: ASSERT(0 <= size);
On 2010/03/04 13:23:15, antonm wrote:
> I am not sure we should ever allocate empty fixed uninitialized arrays.

We won't. Heap::Allocate... returns a singleton in this case. Or do you think
it's error even to request such an array?

[antonm, 2010-03-04 13:28:21]

On 2010/03/04 13:25:58, Vitaly wrote:
> http://codereview.chromium.org/669060/diff/1/3
> File src/factory.cc (right):
> 
> http://codereview.chromium.org/669060/diff/1/3#newcode47
> src/factory.cc:47: ASSERT(0 <= size);
> On 2010/03/04 13:23:15, antonm wrote:
> > I am not sure we should ever allocate empty fixed uninitialized arrays.
> 
> We won't. Heap::Allocate... returns a singleton in this case. Or do you think
> it's error even to request such an array?

Not error, but it looks suspicious for me.  For example, in the optimized cases
(those interested in uninitialized arrays), zero size array usually means a lot
of shortcuts could be taken and there is no need to allocate empty array.  If
you really need one, use Heap::

But I am not insisting.

[Mads Ager (chromium), 2010-03-04 13:30:56]

LGTM

http://codereview.chromium.org/669060/diff/1/7
File src/runtime.cc (right):

http://codereview.chromium.org/669060/diff/1/7#newcode4256
src/runtime.cc:4256: UPDATE_WRITE_BARRIER);
Just omit the WriteBarrierMode here since the default is to update the write
barrier.  The same for the two occurrences below.

[Vitaly Repeshko, 2010-03-04 14:03:11]

http://codereview.chromium.org/669060/diff/1/7
File src/runtime.cc (right):

http://codereview.chromium.org/669060/diff/1/7#newcode4256
src/runtime.cc:4256: UPDATE_WRITE_BARRIER);
On 2010/03/04 13:30:56, Mads Ager wrote:
> Just omit the WriteBarrierMode here since the default is to update the write
> barrier.  The same for the two occurrences below.

Done.