Fixed a bug exposed by the Mozilla test suite where the result of ...

Fixed a bug exposed by the Mozilla test suite where the result of 
a store to a global property is wrong.
"Boolean(x=0)" is one test case.


Committed: http://code.google.com/p/v8/source/detail?r=1446

[Kasper Lund, 2009-03-09 09:27:33]

LGTM. Do we want a simple regression test case too?

http://codereview.chromium.org/40291/diff/1/2
File src/objects.cc (right):

http://codereview.chromium.org/40291/diff/1/2#newcode438
Line 438: Object* store_value = value;
Maybe move the declaration of store_value to the (entry == -1) case?

http://codereview.chromium.org/40291/diff/1/2#newcode443
Line 443: if (store_value->IsFailure()) return store_value;
Move inside the IsJSGlobalObject() case?

http://codereview.chromium.org/40291/diff/1/2#newcode458
Line 458: store_value = cell;
Remove. You're not going to use store_value from here.

http://codereview.chromium.org/40291/diff/1/2#newcode461
Line 461: property_dictionary()->SetStringEntry(entry, name, store_value,
details);
store_value -> value?

http://codereview.chromium.org/40291/diff/1/2#newcode1337
Line 1337: set_properties(Dictionary::cast(result));
Weird indentation.