Make ClassMirror.constructors an internal native and lazy.

Make ClassMirror.constructors an internal native and lazy.

R=asiva@google.com

Committed: https://code.google.com/p/dart/source/detail?r=25533

[rmacnak, 2013-07-26 00:44:28]

https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc
File runtime/lib/mirrors.cc (right):

https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
runtime/lib/mirrors.cc:980: const Error& error =
Error::Handle(klass.EnsureIsFinalized(isolate));
If one reflects on the class before calling any of its members, as many of the
tests do, without finalization it claims to have no members. Previously, the API
version of the constructors map ensured the class was finalized.

[siva, 2013-07-26 17:16:00]

https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc
File runtime/lib/mirrors.cc (right):

https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
runtime/lib/mirrors.cc:1037: // should not wind up as the reflectees of
ClassMirrors.
Can you file a bug for the case of not being able to initial 'dynamic' with an
empty_array and assign it to me. We shouldmake this code uniform and not have to
worry about these special types.

https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
runtime/lib/mirrors.cc:1043: GrowableObjectArray::New(num_functions));
I still think this should be a plain array and not a growable array.
You are not getting any advantage by making it a growable array (just more
memory overhead and an extra deref on every access).

[rmacnak, 2013-07-26 17:42:45]

On 2013/07/26 17:16:00, siva wrote:
> https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc
> File runtime/lib/mirrors.cc (right):
> 
>
https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
> runtime/lib/mirrors.cc:1037: // should not wind up as the reflectees of
> ClassMirrors.
> Can you file a bug for the case of not being able to initial 'dynamic' with an
> empty_array and assign it to me. We shouldmake this code uniform and not have
to
> worry about these special types.

Issue 12071.

>
https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
> runtime/lib/mirrors.cc:1043: GrowableObjectArray::New(num_functions));
> I still think this should be a plain array and not a growable array.
> You are not getting any advantage by making it a growable array (just more
> memory overhead and an extra deref on every access).

Converted to a plain array and adjusted makeMemberMap to deal with gaps.

[rmacnak, 2013-07-26 17:56:59]

On 2013/07/26 17:42:45, Ryan Macnak wrote:
> On 2013/07/26 17:16:00, siva wrote:
> > https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc
> > File runtime/lib/mirrors.cc (right):
> > 
> >
>
https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
> > runtime/lib/mirrors.cc:1037: // should not wind up as the reflectees of
> > ClassMirrors.
> > Can you file a bug for the case of not being able to initial 'dynamic' with
an
> > empty_array and assign it to me. We shouldmake this code uniform and not
have
> to
> > worry about these special types.
> 
> Issue 12071.
> 
> >
>
https://codereview.chromium.org/20564002/diff/1/runtime/lib/mirrors.cc#newcod...
> > runtime/lib/mirrors.cc:1043: GrowableObjectArray::New(num_functions));
> > I still think this should be a plain array and not a growable array.
> > You are not getting any advantage by making it a growable array (just more
> > memory overhead and an extra deref on every access).
> 
> Converted to a plain array and adjusted makeMemberMap to deal with gaps.

As discussed offline, reverted to growable arrays.

[siva, 2013-07-26 18:18:23]

lgtm

[rmacnak, 2013-07-26 18:29:42]

Committed patchset #3 manually as r25533 (presubmit successful).