Use Object::null_instance and Object::null_function and do not create Handles everytime.

Use Object::null_instance and Object::null_function and do not create Handles everytime.

BUG=
R=hausner@google.com

Committed: https://github.com/dart-lang/sdk/commit/228e0877a7692332404ec9a20b2e2fe2ec134b54

[siva, 2015-08-13 01:57:52]

asiva@google.com changed reviewers:
+ hausner@google.com

[srdjan, 2015-08-13 15:49:00]

srdjan@google.com changed reviewers:
+ srdjan@google.com

[srdjan, 2015-08-13 15:49:01]

DBC

https://codereview.chromium.org/1289813005/diff/1/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/1289813005/diff/1/runtime/vm/object.cc#newcod...
runtime/vm/object.cc:7257: result.set_initializer(Object::null_function());
Is this necessary at all, since all data is initialized to Object::null() at
allocation.

[hausner, 2015-08-13 16:09:55]

LGTM. Was this an issue with performance? These handles weren't created at a
high rate. I'm more concerned with the additional field in RawField, now that
I'm aware of it. But that's a different discussion.

[siva, 2015-08-13 16:51:58]

On 2015/08/13 16:09:55, hausner wrote:
> LGTM. Was this an issue with performance? These handles weren't created at a
> high rate. I'm more concerned with the additional field in RawField, now that
> I'm aware of it. But that's a different discussion.

I did not measure any performance impact, Whenever I see wasted creation of
handles I feel I need to get rid of it right away.

[siva, 2015-08-13 16:52:02]

https://codereview.chromium.org/1289813005/diff/1/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/1289813005/diff/1/runtime/vm/object.cc#newcod...
runtime/vm/object.cc:7257: result.set_initializer(Object::null_function());
On 2015/08/13 15:49:01, srdjan wrote:
> Is this necessary at all, since all data is initialized to Object::null() at
> allocation.

That is a good point, not sure why it was there in the first place, I should
other ::New functions to see if we have a pattern of people initializing with
null object.

[siva, 2015-08-13 17:57:22]

Committed patchset #2 (id:20001) manually as
228e0877a7692332404ec9a20b2e2fe2ec134b54 (presubmit successful).

[Florian Schneider, 2015-08-14 09:02:22]

On 2015/08/13 16:51:58, siva wrote:
> On 2015/08/13 16:09:55, hausner wrote:
> > LGTM. Was this an issue with performance? These handles weren't created at a
> > high rate. I'm more concerned with the additional field in RawField, now
that
> > I'm aware of it. But that's a different discussion.
> 
> I did not measure any performance impact, Whenever I see wasted creation of
> handles I feel I need to get rid of it right away.

+1!