Templatise type representation

Templatise type representation

This is to support both heap- and zone-allocated types in the future (the latter not yet implemented).

Also, handlify the type API some more.

R=titzer@chromium.org
BUG=

Committed: https://code.google.com/p/v8/source/detail?r=18521

[titzer, 2013-12-17 09:13:07]

lgtm,

Pretty amazing wizardry, thanks C++!

https://codereview.chromium.org/107933005/diff/1/src/ic.cc
File src/ic.cc (right):

https://codereview.chromium.org/107933005/diff/1/src/ic.cc#newcode1012
src/ic.cc:1012: ? Type::Constant(Handle<JSGlobalObject>::cast(object), isolate)
Side question: It's weird the type of the global object is itself. Why?

https://codereview.chromium.org/107933005/diff/1/src/list.h
File src/list.h (right):

https://codereview.chromium.org/107933005/diff/1/src/list.h#newcode212
src/list.h:212: typedef List<Map*> MapList;
I find it weird that these typedefs are already in the general header. Why?

https://codereview.chromium.org/107933005/diff/1/src/objects.h
File src/objects.h (right):

https://codereview.chromium.org/107933005/diff/1/src/objects.h#newcode884
src/objects.h:884: // C++ is a pile of garbage with no proper compositionality
or abstraction.
Agree with the sentiment--and glad to have a comment explaining why-- but
vitriol should be limited to code review comments, you ____-_______ing ____er :)

https://codereview.chromium.org/107933005/diff/1/src/property-details.h
File src/property-details.h (right):

https://codereview.chromium.org/107933005/diff/1/src/property-details.h#newco...
src/property-details.h:60: typedef TypeImpl<HeapTypeConfig> Type;
Oh, it spreads. Love.

[rossberg, 2013-12-17 09:33:13]

https://codereview.chromium.org/107933005/diff/1/src/ic.cc
File src/ic.cc (right):

https://codereview.chromium.org/107933005/diff/1/src/ic.cc#newcode1012
src/ic.cc:1012: ? Type::Constant(Handle<JSGlobalObject>::cast(object), isolate)
On 2013/12/17 09:13:07, titzer wrote:
> Side question: It's weird the type of the global object is itself. Why?

The global object is handled rather specially in the ICs, and you need to know
its identity. Toon can explain the details.

https://codereview.chromium.org/107933005/diff/1/src/list.h
File src/list.h (right):

https://codereview.chromium.org/107933005/diff/1/src/list.h#newcode212
src/list.h:212: typedef List<Map*> MapList;
On 2013/12/17 09:13:07, titzer wrote:
> I find it weird that these typedefs are already in the general header. Why?

I don't think that they belong here either. Care to clean up? ;)

https://codereview.chromium.org/107933005/diff/1/src/objects.h
File src/objects.h (right):

https://codereview.chromium.org/107933005/diff/1/src/objects.h#newcode884
src/objects.h:884: // C++ is a pile of garbage with no proper compositionality
or abstraction.
On 2013/12/17 09:13:07, titzer wrote:
> Agree with the sentiment--and glad to have a comment explaining why-- but
> vitriol should be limited to code review comments, you ____-_______ing ____er
:)

Had to vent. I literally wasted a day trying to find a way that avoids this
abstraction leak all over the place. AFAICT, there is none.

Removed the latter half of the comment.

[rossberg, 2014-01-09 17:13:04]

Committed patchset #1 manually as r18521 (presubmit successful).