Make HTMLParserScheduler inherit from ActiveDOMObject

Make HTMLParserScheduler inherit from ActiveDOMObject

ActiveDOMObject is a standard way to observe suspend/resume.
ScriptRunner should follow the rule. The goal is to remove
ExecutionContext::tasksWereSuspended/Resumed in a follow-up CL.

BUG=610176

[haraken, 2016-12-06 06:42:45]

haraken@chromium.org changed reviewers:
+ sigbjornf@opera.com

[haraken, 2016-12-06 06:42:46]

PTAL

[sof, 2016-12-06 08:20:05]

sigbjornf@opera.com changed reviewers:
+ dcheng@chromium.org, kouhei@chromium.org

[sof, 2016-12-06 08:20:05]

Adding some reviewers; the intrinsic value of this push escapes me somewhat.

[sof, 2016-12-07 12:26:51]

On 2016/12/06 08:20:05, sof wrote:
> Adding some reviewers; the intrinsic value of this push escapes me somewhat.

(iow, i don't want to be standing in the way for anyone else wanting to approve
it.)

[haraken, 2016-12-07 12:53:22]

On 2016/12/07 12:26:51, sof wrote:
> On 2016/12/06 08:20:05, sof wrote:
> > Adding some reviewers; the intrinsic value of this push escapes me somewhat.
> 
> (iow, i don't want to be standing in the way for anyone else wanting to
approve
> it.)

I don't have any strong opinion on this CL since it's not blocking anything -- I
just found that ExecutionContext::tasksWereSuspended is redundant and wanted to
remove it. If you want to keep it, I'm okay with it. Not planning to spend
energy on this :)

[dcheng, 2016-12-08 08:33:07]

On one hand, it feels a little odd for the HTMLParserScheduler to be an
ActiveDOMObject: when I see a subclass, I usually think that it's something
exposed to DOM/JS.

However, it's also nice not to have a one-off interface just for one thing... on
the whole, I think I lean in favor of this patch.

LGTM, though I'm a bit uncertain about the naming here...

[haraken, 2016-12-08 08:46:03]

On 2016/12/08 08:33:07, dcheng wrote:
> On one hand, it feels a little odd for the HTMLParserScheduler to be an
> ActiveDOMObject: when I see a subclass, I usually think that it's something
> exposed to DOM/JS.
> 
> However, it's also nice not to have a one-off interface just for one thing...
on
> the whole, I think I lean in favor of this patch.
> 
> LGTM, though I'm a bit uncertain about the naming here...

I think ActiveDOMObject should be renamed to SuspendableObject. It's an
abstraction that provides suspend/resume.

[dcheng, 2016-12-08 08:46:51]

On 2016/12/08 08:46:03, haraken wrote:
> On 2016/12/08 08:33:07, dcheng wrote:
> > On one hand, it feels a little odd for the HTMLParserScheduler to be an
> > ActiveDOMObject: when I see a subclass, I usually think that it's something
> > exposed to DOM/JS.
> > 
> > However, it's also nice not to have a one-off interface just for one
thing...
> on
> > the whole, I think I lean in favor of this patch.
> > 
> > LGTM, though I'm a bit uncertain about the naming here...
> 
> I think ActiveDOMObject should be renamed to SuspendableObject. It's an
> abstraction that provides suspend/resume.

That makes sense to me. LGTM

[kouhei (in TOK), 2016-12-08 11:22:01]

lgtm given the rename plan