Introduce Unsigned32 and RegExp types

Introduce Unsigned32 and RegExp types

Also, renamed Integer31 back to Smi, and Integer32 to Signed32.

R=jkummerow@chromium.org
BUG=

Committed: http://code.google.com/p/v8/source/detail?r=15332

[Jakob Kummerow, 2013-06-24 15:12:11]

LGTM with a nit.

https://codereview.chromium.org/17589013/diff/1/src/types.h
File src/types.h (right):

https://codereview.chromium.org/17589013/diff/1/src/types.h#newcode52
src/types.h:52: //   Smi < Signed32
strictly speaking: Smi <= Signed32 (as well as Signed31 <= Smi for a
hypothetical Signed31 type).

[Vyacheslav Egorov (Google), 2013-06-24 15:14:16]

DBC: Smi on x64 is a Signed32. How does it fit with naming?

[Jakob Kummerow, 2013-06-24 15:18:52]

On 2013/06/24 15:14:16, Vyacheslav Egorov (Google) wrote:
> DBC: Smi on x64 is a Signed32. How does it fit with naming?

That's the point of the Integer31 -> Smi renaming. The idea is that "Smi" is
platform agnostic, it just means "whatever can be stored as a Smi on this
platform". It should play well with existing code.
(See my comment that the relation is really "Smi <= Signed32".)

[rossberg, 2013-06-25 06:21:51]

https://codereview.chromium.org/17589013/diff/1/src/types.h
File src/types.h (right):

https://codereview.chromium.org/17589013/diff/1/src/types.h#newcode52
src/types.h:52: //   Smi < Signed32
On 2013/06/24 15:12:11, Jakob wrote:
> strictly speaking: Smi <= Signed32 (as well as Signed31 <= Smi for a
> hypothetical Signed31 type).

Good point.

[rossberg, 2013-06-26 09:52:05]

Committed patchset #2 manually as r15332 (presubmit successful).