Refactor the class id generation code a bit to make it possible to have toActiveDOMObject, ToEventT…

Refactor the class id generation code a bit to make it possible to have toActiveDOMObject, ToEventTarget and toNode functions for each webkit type.

R=rmacnak@google.com

Committed: https://code.google.com/p/dart/source/detail?r=34840

[rmacnak, 2014-04-05 00:37:31]

lgtm

[siva, 2014-04-08 20:28:02]

Committed patchset #2 manually as r34840 (presubmit successful).

[vsm, 2014-04-08 20:29:27]

lgtm

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
File tools/dom/scripts/systemnative.py (right):

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
tools/dom/scripts/systemnative.py:628: else:
Should these generate code to assert unreachable in the else cases?

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
tools/dom/scripts/systemnative.py:1449: is_event_target_test = lambda interface:
'EventTarget' in interface.ext_attrs
This may need to be lambda interface: interface.id == 'EventTarget'.  Not
positive, but I don't think it's an attribute any more.

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
tools/dom/scripts/systemnative.py:1526: e.Emit('ActiveDOMObject*
toNullActiveDOMObject(void* value) { return 0; }\n');
Ditto on ASSERTs here - will these ever be called?

https://codereview.chromium.org/224993003/diff/20001/tools/dom/templates/html...
File tools/dom/templates/html/dartium/cpp_header.template (right):

https://codereview.chromium.org/224993003/diff/20001/tools/dom/templates/html...
tools/dom/templates/html/dartium/cpp_header.template:27: static ActiveDOMObject*
toActiveDOMObject(void* value)
if HasConverters is false for the type, should these be suppressed altogether? 
it looks like they are never referenced in that case.

[siva, 2014-04-08 23:09:03]

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
File tools/dom/scripts/systemnative.py (right):

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
tools/dom/scripts/systemnative.py:628: else:
When iterating over the weak handles we would use the return 0 to ensure that we
don't try to process something that is not a Node for instance. You will see the
code in my next CL.

On 2014/04/08 20:29:27, vsm wrote:
> Should these generate code to assert unreachable in the else cases?

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
tools/dom/scripts/systemnative.py:1449: is_event_target_test = lambda interface:
'EventTarget' in interface.ext_attrs
I will fix this and another use of is_event_target below in a new CL.


On 2014/04/08 20:29:27, vsm wrote:
> This may need to be lambda interface: interface.id == 'EventTarget'.  Not
> positive, but I don't think it's an attribute any more.

https://codereview.chromium.org/224993003/diff/20001/tools/dom/scripts/system...
tools/dom/scripts/systemnative.py:1526: e.Emit('ActiveDOMObject*
toNullActiveDOMObject(void* value) { return 0; }\n');
Yes they would be used to trigger processing in the GC prologue code.

On 2014/04/08 20:29:27, vsm wrote:
> Ditto on ASSERTs here - will these ever be called?

https://codereview.chromium.org/224993003/diff/20001/tools/dom/templates/html...
File tools/dom/templates/html/dartium/cpp_header.template (right):

https://codereview.chromium.org/224993003/diff/20001/tools/dom/templates/html...
tools/dom/templates/html/dartium/cpp_header.template:27: static ActiveDOMObject*
toActiveDOMObject(void* value)
I don't yet have the GC prolog code done yet. You will see in that code that we
use these methods to figure out if an object needs to be processed as a Node,
Event target or an Active object.

On 2014/04/08 20:29:27, vsm wrote:
> if HasConverters is false for the type, should these be suppressed altogether?

> it looks like they are never referenced in that case.