Put metadata fields in their owner class.

Put metadata fields in their owner class to maintain the class-field bidirectional relationship asserted Field::PrintJSON.

Extended metadata scope test based on suspicions regarding the ordering of adding type parameters metadata fields versus ordinary fields of a class.

BUG=http://dartbug.com/18652
R=iposva@google.com, regis@google.com

Committed: https://code.google.com/p/dart/source/detail?r=35926

[rmacnak, 2014-05-08 01:35:21]

https://codereview.chromium.org/269253005/diff/40001/runtime/vm/class_finaliz...
File runtime/vm/class_finalizer.cc (left):

https://codereview.chromium.org/269253005/diff/40001/runtime/vm/class_finaliz...
runtime/vm/class_finalizer.cc:2292: ASSERT(Array::Handle(cls.fields()).Length()
== 0);
This assert no longer holds because we can have metadata before the class is
fully compiled.  Put a bit in Field marking it as a metadata field and assert
only metadata fields here?

https://codereview.chromium.org/269253005/diff/40001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/269253005/diff/40001/runtime/vm/object.cc#new...
runtime/vm/object.cc:2797: void Class::AddField(const Field& field) const {
Cf. AddFunction

[Ivan Posva, 2014-05-08 05:18:49]

LGTM -ip

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/class_finaliz...
File runtime/vm/class_finalizer.cc (right):

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/class_finaliz...
runtime/vm/class_finalizer.cc:2288: // This class should not contain any fields
or functions yet, because it has
Please update the comment. Especially since it appears that this class now can
already contain fields when we get here.

[regis, 2014-05-08 16:39:10]

LGTM

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/object.cc#new...
runtime/vm/object.cc:2799: const Array& new_arr = Array::Handle(Array::Grow(arr,
arr.Length() + 1));
By default, Array::Grow will allocate in the new space. Is it what you want? I
do not know how frequently you add a field.

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/object.cc#new...
runtime/vm/object.cc:2811: Array::Handle(Array::Grow(arr, num_old_fields +
num_new_fields));
ditto

[rmacnak, 2014-05-08 18:14:54]

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/class_finaliz...
File runtime/vm/class_finalizer.cc (right):

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/class_finaliz...
runtime/vm/class_finalizer.cc:2288: // This class should not contain any fields
or functions yet, because it has
On 2014/05/08 05:18:50, Ivan Posva wrote:
> Please update the comment. Especially since it appears that this class now can
> already contain fields when we get here.

Noted the class might contain metadata fields for type parameters.

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/object.cc#new...
runtime/vm/object.cc:2799: const Array& new_arr = Array::Handle(Array::Grow(arr,
arr.Length() + 1));
On 2014/05/08 16:39:11, regis wrote:
> By default, Array::Grow will allocate in the new space. Is it what you want? I
> do not know how frequently you add a field.

This path will only be hit when we encounter metadata.

https://codereview.chromium.org/269253005/diff/60001/runtime/vm/object.cc#new...
runtime/vm/object.cc:2811: Array::Handle(Array::Grow(arr, num_old_fields +
num_new_fields));
On 2014/05/08 16:39:11, regis wrote:
> ditto

This will happen once when we parse the class members / clone a mixin.

[rmacnak, 2014-05-08 18:15:40]

Committed patchset #4 manually as r35926 (presubmit successful).