Current HandleScope moving to Isolate.

Current HandleScope moving to Isolate.
This is first part. HandleScopeImplementor is still global.

Committed: http://code.google.com/p/v8/source/detail?r=4788

[Vitaly Repeshko, 2010-06-02 08:54:21]

LGTM

http://codereview.chromium.org/2476001/diff/1/2
File src/api.cc (right):

http://codereview.chromium.org/2476001/diff/1/2#newcode446
src/api.cc:446: EnsureInitialized("HandleScope::HandleScope");
Please add comment describing why this is required.

http://codereview.chromium.org/2476001/diff/1/2#newcode3052
src/api.cc:3052: // Make sure the initialization did not leave allocated handles
in
Does NoHandleAllocation work here?

http://codereview.chromium.org/2476001/diff/1/5
File src/handles.cc (right):

http://codereview.chromium.org/2476001/diff/1/5#newcode58
src/handles.cc:58: Isolate::Current()->handle_scope_data();
Ditto.

http://codereview.chromium.org/2476001/diff/1/5#newcode98
src/handles.cc:98: Isolate::Current()->handle_scope_data();
Ditto.

http://codereview.chromium.org/2476001/diff/1/6
File src/handles.h (right):

http://codereview.chromium.org/2476001/diff/1/6#newcode125
src/handles.h:125: Isolate::Current()->handle_scope_data();
nit: 4 space indent on line continuations.

http://codereview.chromium.org/2476001/diff/1/6#newcode159
src/handles.h:159: Isolate::Current()->handle_scope_data();
Ditto.

http://codereview.chromium.org/2476001/diff/1/6#newcode169
src/handles.h:169: Isolate::Current()->handle_scope_data();
Ditto.

[Dmitry Titov, 2010-06-02 11:40:25]

All fixed, and I additionally modified tests in test-serialize.
They used the sequence:

HandleScope scope;
Deserialize(); // from test file

which was not reading the file at all because HandleScope would already
initialize the default isolate, so the attempt to initialize it again by
deserializer will just immediately return. Moved HandleScope constructor below
Initialize() call, or removed it at all if no handles are created in the test.
Since we don't expose deserialization in API it seems ok.

> http://codereview.chromium.org/2476001/diff/1/2#newcode3052
> src/api.cc:3052: // Make sure the initialization did not leave allocated
handles
> in
> Does NoHandleAllocation work here?

It wouldn't work here because first thing its constructor does is to create a
HandleScope again, before current isolate is initialized. But even better - we
don't need any checks because there is already one in HandleScope::Extend that
would call ReportApiFailure(...) if there is no scope on the stack on attempt to
create a handle.

[Vitaly Repeshko, 2010-06-02 15:04:36]

Still LGTM.