DOM fixes for types that are not marked hidden

DOM fixes for types that are not marked hidden. They are not used, so we don't generate the right dynamic handlers for them.

Committed: https://code.google.com/p/dart/source/detail?r=2283

[Jennifer Messerly, 2011-12-08 21:47:17]

Let me know how if this looks okay to checkin. Will unblock some people trying
to use things like Event.preventDefault

[Jennifer Messerly, 2011-12-08 21:57:12]

On 2011/12/08 21:47:17, John Messerly wrote:
> Let me know how if this looks okay to checkin. Will unblock some people trying
> to use things like Event.preventDefault

submitting tbr

[sra1, 2011-12-08 22:04:28]

http://codereview.chromium.org/8888027/diff/1/client/dom/frog/frog_dom.dart
File client/dom/frog/frog_dom.dart (right):

http://codereview.chromium.org/8888027/diff/1/client/dom/frog/frog_dom.dart#n...
client/dom/frog/frog_dom.dart:3222: class HTMLMediaElement extends HTMLElement
native "*HTMLMediaElement" {
This will likely break FireFox since it is not a leaf.

Does HTMLMediaElement not exist?  The list of non-star types is taken by
intersecting available names on a set of browsers with the non-leaves.

Plus some exceptions to work around I bug I need to file.

http://codereview.chromium.org/8888027/diff/1/frog/gen.dart
File frog/gen.dart (right):

http://codereview.chromium.org/8888027/diff/1/frog/gen.dart#newcode168
frog/gen.dart:168: if ((type.isUsed || type.library == world.dom
I assume isUsed get set when a subclass is used?

Then there is a work-around for Event:  
  if (false) { <create a subclass of event> }

But that is harder to do in the source than first appears.

Could you do a pre-walk over the types, marking *Foo types and their parents as
used?

[Jennifer Messerly, 2011-12-08 22:17:47]

http://codereview.chromium.org/8888027/diff/1/client/dom/frog/frog_dom.dart
File client/dom/frog/frog_dom.dart (right):

http://codereview.chromium.org/8888027/diff/1/client/dom/frog/frog_dom.dart#n...
client/dom/frog/frog_dom.dart:3222: class HTMLMediaElement extends HTMLElement
native "*HTMLMediaElement" {
On 2011/12/08 22:04:32, sra1 wrote:
> This will likely break FireFox since it is not a leaf.
> 
> Does HTMLMediaElement not exist?  The list of non-star types is taken by
> intersecting available names on a set of browsers with the non-leaves.
> 
> Plus some exceptions to work around I bug I need to file.

As the comment says, it doesn't exist on DumpRenderTree on Mac. I'd love to have
a workaround. But otherwise I would've broken tests on Mac. Siggi is looking for
a fix--he thinks there is a way we can make DumpRenderTree include this.

I don't understand the FF issue--did you manage to mark all non-leaf types
"visible"? And all those types are visible cross browser? If that's really true,
it's good to know!

http://codereview.chromium.org/8888027/diff/1/frog/gen.dart
File frog/gen.dart (right):

http://codereview.chromium.org/8888027/diff/1/frog/gen.dart#newcode168
frog/gen.dart:168: if ((type.isUsed || type.library == world.dom
On 2011/12/08 22:04:32, sra1 wrote:
> I assume isUsed get set when a subclass is used?
> 
> Then there is a work-around for Event:  
>   if (false) { <create a subclass of event> }
> 
> But that is harder to do in the source than first appears.
> 
> Could you do a pre-walk over the types, marking *Foo types and their parents
as
> used?

Maybe. The "hidden" bit isn't really related, though--that was just a quick hack
to make the DOM work and to make your "makeA" tests work.

What we need is to be more explicit--for every native method, including corelib,
coreimpl, DOM and any special Frog tests, we need the list of types it might
create--similar to what we do some places in corelib. (the block of code
following the "native" declaration).