Change interface to CodeGenerator::LoadFromSlot.

Change interface to CodeGenerator::LoadFromSlot.

Change the interface to LoadFromSlot, LoadFromSlotCheckForArguments,
and LoadFromGlobalSlotCheckExtensions in the ia32 classic code
generator.

Committed: http://code.google.com/p/v8/source/detail?r=3849

[fschneider, 2010-02-12 12:23:05]

LGTM. 

Will you include the change for the x64 back-end? I think we should keep them in
sync.

http://codereview.chromium.org/606015/diff/1/2
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/606015/diff/1/2#newcode3997
src/ia32/codegen-ia32.cc:3997: Result result;
I'd move the declaration into the innermost necessary scope because it is not
used on some paths.

http://codereview.chromium.org/606015/diff/1/2#newcode3999
src/ia32/codegen-ia32.cc:3999: ASSERT(slot->var()->is_dynamic());
Declare result here instead.

http://codereview.chromium.org/606015/diff/1/2#newcode4015
src/ia32/codegen-ia32.cc:4015: Slot* potential_slot =
slot->var()->local_if_not_shadowed()->slot();
Declare result here.

http://codereview.chromium.org/606015/diff/1/2#newcode4076
src/ia32/codegen-ia32.cc:4076: return frame()->Pop();
Would it make sense to have a virtual frame function 
"Result getParameterAt(index)" or 
"Result getLocalAt(index)" instead of pushing/popping?

http://codereview.chromium.org/606015/diff/1/2#newcode4083
src/ia32/codegen-ia32.cc:4083: // The other remaining slot types (LOOKUP and
GLOBAL) cannot reach
Declare result here.

[Kevin Millikin (Chromium), 2010-02-12 12:39:21]

http://codereview.chromium.org/606015/diff/1/2
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/606015/diff/1/2#newcode3997
src/ia32/codegen-ia32.cc:3997: Result result;
On 2010/02/12 12:23:06, fschneider wrote:
> I'd move the declaration into the innermost necessary scope because it is not
> used on some paths.

Ultimately it will be a field in the codegen state, so I really want one
declaration at the very top of the function (so it's easy to remove) called
result (so it's easy to search for).

http://codereview.chromium.org/606015/diff/1/2#newcode4076
src/ia32/codegen-ia32.cc:4076: return frame()->Pop();
On 2010/02/12 12:23:06, fschneider wrote:
> Would it make sense to have a virtual frame function 
> "Result getParameterAt(index)" or 
> "Result getLocalAt(index)" instead of pushing/popping?

Yes, and PushParameterAt should go away.  That's a separate change I'll make
after it's no longer used.