Fixed numeric relations on HPhi instances.

Fixed numeric relations on HPhi instances.

Committed: https://code.google.com/p/v8/source/detail?r=13703

[Jakob Kummerow, 2013-02-20 10:46:47]

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h
File src/hydrogen-instructions.h (right):

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h#n...
src/hydrogen-instructions.h:708: // If a phi is involved in the evaluation of a
numeric constraints the
"a numeric constraints" -> singular or plural?

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h#n...
src/hydrogen-instructions.h:2969: class HInductionVariable : public
HUnaryOperation {
The name is misleading, as this instruction *isn't* an induction variable, it's
just an idef. Maybe "InductionVariableAnnotation"?

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h#n...
src/hydrogen-instructions.h:3004: HPhi* phi_;
No need to store this -- the UnaryOperation base class stores it for us as
"value()".

[Massi, 2013-02-21 07:40:32]

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h
File src/hydrogen-instructions.h (right):

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h#n...
src/hydrogen-instructions.h:708: // If a phi is involved in the evaluation of a
numeric constraints the
On 2013/02/20 10:46:47, Jakob wrote:
> "a numeric constraints" -> singular or plural?

Done.

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h#n...
src/hydrogen-instructions.h:2969: class HInductionVariable : public
HUnaryOperation {
On 2013/02/20 10:46:47, Jakob wrote:
> The name is misleading, as this instruction *isn't* an induction variable,
it's
> just an idef. Maybe "InductionVariableAnnotation"?

Done.

https://codereview.chromium.org/12301027/diff/1/src/hydrogen-instructions.h#n...
src/hydrogen-instructions.h:3004: HPhi* phi_;
On 2013/02/20 10:46:47, Jakob wrote:
> No need to store this -- the UnaryOperation base class stores it for us as
> "value()".

Actually it is needed, adding a comment to clarify why.

[Jakob Kummerow, 2013-02-21 10:16:37]

lgtm

[Massi, 2013-02-21 10:22:39]

Committed manually as r13703 (presubmit successful).