Set instruction/use_index when adding an input to an IL instruction.

Set instruction/use_index when adding an input to an IL instruction.

When setting an input (including in the constructor) of an IL instruction,
automatically set the input use's instruction and use_index fields.

R=vegorov@google.com

Committed: https://code.google.com/p/dart/source/detail?r=18961

[Vyacheslav Egorov (Google), 2013-02-22 16:13:36]

lgtm

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
File runtime/vm/intermediate_language.cc (right):

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language.cc:572: for (intptr_t i =
comparison->InputCount() - 1; i >= 0; --i) {
Why are we going backwards?

Seems like a micro-optimization.

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language.cc:579: comparison()->SetInputAt(i, value);
I think here you should delegate to RawSetInputAt.

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
File runtime/vm/intermediate_language.h (right):

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language.h:1420: friend class ConstantPropagator;
Comment does not match implementation. ConstantPropagator seems to call
SetInputAt.

[Kevin Millikin (Google), 2013-02-25 11:08:07]

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
File runtime/vm/intermediate_language.cc (right):

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language.cc:572: for (intptr_t i =
comparison->InputCount() - 1; i >= 0; --i) {
On 2013/02/22 16:13:36, Vyacheslav Egorov (Google) wrote:
> Why are we going backwards?
> 
> Seems like a micro-optimization.

I disagree.

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language.cc:579: comparison()->SetInputAt(i, value);
On 2013/02/22 16:13:36, Vyacheslav Egorov (Google) wrote:
> I think here you should delegate to RawSetInputAt.

Oops, OK.

It's always seemed arbitrary (and more complicated) to me that the branch steals
ownership of the inputs from the comparison.  That the code above works suggests
that nobody depends on the current behavior.

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
File runtime/vm/intermediate_language.h (right):

https://codereview.chromium.org/12316065/diff/1/runtime/vm/intermediate_langu...
runtime/vm/intermediate_language.h:1420: friend class ConstantPropagator;
On 2013/02/22 16:13:36, Vyacheslav Egorov (Google) wrote:
> Comment does not match implementation. ConstantPropagator seems to call
> SetInputAt.

It uses friend access to call inputs_.Truncate(...) to resize inputs_.  Any
suggestion to make the comment more clear?

[Kevin Millikin (Google), 2013-02-25 11:24:17]

Committed patchset #2 manually as r18961 (presubmit successful).