Change the representation of catch contexts.

Change the representation of catch contexts.

Before, they had no extra slots and an extension object with one named
property.  Now, they use the extension slot for the property name and have
an extra slot for the thrown object.  This increases the size of the context
itself, but removes overall allocation and eliminates a level of indirection.

R=ager@chromium.org

Committed: http://code.google.com/p/v8/source/detail?r=8277

[Mads Ager (chromium), 2011-06-14 11:54:17]

LGTM

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc#newcode10188
src/runtime.cc:10188:
isolate->factory()->NewJSObject(isolate->object_function());
Do we really want this to be a plain JSObject. Don't we need this to be a
context extension object so we do not start looking up the prototype chain? I
guess this case is handled specifically in contexts.cc so it will be OK?

[Kevin Millikin (Chromium), 2011-06-14 12:06:35]

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc#newcode10188
src/runtime.cc:10188:
isolate->factory()->NewJSObject(isolate->object_function());
On 2011/06/14 11:54:17, Mads Ager wrote:
> Do we really want this to be a plain JSObject. Don't we need this to be a
> context extension object so we do not start looking up the prototype chain? I
> guess this case is handled specifically in contexts.cc so it will be OK?

I'm pretty sure this is what we want.  This is an object that's put on an
indexed property of a "scope details" mirror object used by the debugger.  I'll
run it by Søren.

[Søren Thygesen Gjesse, 2011-06-14 12:18:32]

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc#newcode10188
src/runtime.cc:10188:
isolate->factory()->NewJSObject(isolate->object_function());
On 2011/06/14 11:54:17, Mads Ager wrote:
> Do we really want this to be a plain JSObject. Don't we need this to be a
> context extension object so we do not start looking up the prototype chain? I
> guess this case is handled specifically in contexts.cc so it will be OK?

It looks fine. The objects representing the different scopes are just plain
read-only JSObjects created on demand.

[Søren Thygesen Gjesse, 2011-06-14 12:18:32]

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/7152002/diff/1/src/runtime.cc#newcode10188
src/runtime.cc:10188:
isolate->factory()->NewJSObject(isolate->object_function());
On 2011/06/14 11:54:17, Mads Ager wrote:
> Do we really want this to be a plain JSObject. Don't we need this to be a
> context extension object so we do not start looking up the prototype chain? I
> guess this case is handled specifically in contexts.cc so it will be OK?

It looks fine. The objects representing the different scopes are just plain
read-only JSObjects created on demand.

[Søren Thygesen Gjesse, 2011-06-14 12:23:31]

http://codereview.chromium.org/7152002/diff/1/src/contexts.cc
File src/contexts.cc (right):

http://codereview.chromium.org/7152002/diff/1/src/contexts.cc#newcode103
src/contexts.cc:103: if (name->Equals(String::cast(context->extension()))) {
Dbc. Should context perhaps have a method GetCatchVariableName?