Devirtualize ExceptionState's helper throwing methods.

Devirtualize ExceptionState's helper throwing methods.

All the helper methods end up delegating to setException(), so only
setException() needs to be virtual. This reduces binary size by 44KB on
a Linux x64 build.

A minor disadvantage is throwing with the subclasses is less efficient
now: an Exception object is created, only to be discarded when calling
setException(). However, throwing an exception shouldn't be on the hot
code path, so the trade off should be acceptable.

BUG=none

[dcheng, 2017-02-25 12:00:33]

dcheng@chromium.org changed reviewers:
+ jbroman@chromium.org

[dcheng, 2017-02-25 12:00:33]

This one may be more controversial. =)

Note that it's possible to get rid of the inefficiency of the ignored Exception
object by devirtualizing setException() and adding a virtual pre-throw hook
instead.

[dcheng, 2017-02-25 12:02:35]

https://codereview.chromium.org/2715073002/diff/1/third_party/WebKit/Source/b...
File third_party/WebKit/Source/bindings/core/v8/ExceptionState.cpp (right):

https://codereview.chromium.org/2715073002/diff/1/third_party/WebKit/Source/b...
third_party/WebKit/Source/bindings/core/v8/ExceptionState.cpp:64: const String&
processedMessage = addExceptionContext(message);
The alternative here is to consult a new policy virtual (shouldThrow?) before
building the exception message.

[dcheng, 2017-02-25 23:39:16]

The CQ bit was checked by dcheng@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-25 23:39:27]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-26 01:00:24]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-26 01:00:24]

Dry run: Try jobs failed on following builders:
  linux_chromium_asan_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[dcheng, 2017-02-26 11:48:06]

Ah... I guess this has to be done with a shouldThrow() virtual, as the
subclasses don't set an isolate. Blah =)

[haraken, 2017-02-26 16:14:27]

On 2017/02/26 11:48:06, dcheng wrote:
> Ah... I guess this has to be done with a shouldThrow() virtual, as the
> subclasses don't set an isolate. Blah =)

So, is this ready for review?

[dcheng, 2017-02-27 10:53:22]

On 2017/02/26 16:14:27, haraken wrote:
> On 2017/02/26 11:48:06, dcheng wrote:
> > Ah... I guess this has to be done with a shouldThrow() virtual, as the
> > subclasses don't set an isolate. Blah =)
> 
> So, is this ready for review?

Sorry, my original approach doesn't work. I didn't realize it originally, but
DummyExceptionStateForTesting is also problematic: the header says it ignores
all thrown exceptions, but for tests, it's still important to correctly track
whether or not an exception was thrown...
It's likely the tests should just be using a real ExceptionState, but I will
need to investigate and see how difficult that will be.

I'll re-ping when this is ready for review.