Make sure the bound of a mixin type parameter is finalized before instantiating

Make sure the bound of a mixin type parameter is finalized before instantiating
it during finalization (fix issue 23385).
Print more info in ToCString() for an unresolved type.

Committed: https://code.google.com/p/dart/source/detail?r=45598

[regis, 2015-05-07 10:26:59]

regis@google.com changed reviewers:
+ iposva@google.com, srdjan@google.com

[regis, 2015-05-07 10:27:00]

It is difficult to write a regression test, because the issue depends in the
order in which types and classes are finalized. The test case reported in the
bug report is now passing successfully.

Thanks,
Regis

[Ivan Posva, 2015-05-07 13:21:26]

https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer.cc
File runtime/vm/class_finalizer.cc (right):

https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer....
runtime/vm/class_finalizer.cc:1691: param.set_bound(param_bound);
Can you please add a comment why this is necessary here? It will be overridden
again on line 1699 further down.

[regis, 2015-05-07 13:33:54]

Thanks

https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer.cc
File runtime/vm/class_finalizer.cc (right):

https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer....
runtime/vm/class_finalizer.cc:1691: param.set_bound(param_bound);
On 2015/05/07 13:21:25, Ivan Posva wrote:
> Can you please add a comment why this is necessary here? It will be overridden
> again on line 1699 further down.

In case the bound is part of a recursive type, it may be reached again (via the
mixin app class type parameters) during the instantiation step.
Added comment.

[regis, 2015-05-07 15:57:29]

On 2015/05/07 13:33:54, regis wrote:
> Thanks
> 
>
https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer.cc
> File runtime/vm/class_finalizer.cc (right):
> 
>
https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer....
> runtime/vm/class_finalizer.cc:1691: param.set_bound(param_bound);
> On 2015/05/07 13:21:25, Ivan Posva wrote:
> > Can you please add a comment why this is necessary here? It will be
overridden
> > again on line 1699 further down.
> 
> In case the bound is part of a recursive type, it may be reached again (via
the
> mixin app class type parameters) during the instantiation step.
> Added comment.

Still waiting for lgtm (and I will be in vacation for 2 weeks after tomorrow).

[regis, 2015-05-07 16:07:42]

On 2015/05/07 15:57:29, regis wrote:
> On 2015/05/07 13:33:54, regis wrote:
> > Thanks
> > 
> >
>
https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer.cc
> > File runtime/vm/class_finalizer.cc (right):
> > 
> >
>
https://codereview.chromium.org/1124253004/diff/1/runtime/vm/class_finalizer....
> > runtime/vm/class_finalizer.cc:1691: param.set_bound(param_bound);
> > On 2015/05/07 13:21:25, Ivan Posva wrote:
> > > Can you please add a comment why this is necessary here? It will be
> overridden
> > > again on line 1699 further down.
> > 
> > In case the bound is part of a recursive type, it may be reached again (via
> the
> > mixin app class type parameters) during the instantiation step.
> > Added comment.
> 
> Still waiting for lgtm (and I will be in vacation for 2 weeks after tomorrow).

Got lgtm from Ivan via chat.

[regis, 2015-05-07 16:08:19]

Committed patchset #2 (id:20001) manually as r45598 (presubmit successful).

[Ivan Posva, 2015-05-07 22:52:33]

Official LGTM now as well...

-Ivan