Ignore default reject handler when looking for reject handlers.

Ignore default reject handler when looking for reject handlers.


LOG=N
BUG=v8:3093
R=aandrey@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=23112

[Yang, 2014-08-12 13:28:35]

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/reject-caug...
File test/mjsunit/es6/reject-caught-by-default-reject-handler.js (right):

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/reject-caug...
test/mjsunit/es6/reject-caught-by-default-reject-handler.js:46: assertEquals(0,
exec_state.frameCount());
Note that this is because the promise is rejected from within native JS code due
to the return value being a rejected promise. Devtools will just ignore this
still, because there is no stack trace. I really have no idea how to solve this.

[aandrey, 2014-08-13 11:03:05]

lgtm

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/reject-caug...
File test/mjsunit/es6/reject-caught-by-default-reject-handler.js (right):

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/reject-caug...
test/mjsunit/es6/reject-caught-by-default-reject-handler.js:69:
assertEquals(["resolve q", "end main", "resolve p", "p0.then"], log);
should there be "p1.then" as well?

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
File test/mjsunit/es6/throw-caught-by-default-reject-handler.js (right):

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
test/mjsunit/es6/throw-caught-by-default-reject-handler.js:50: // p2 is rejected
by p1's default reject handler.
do we want the debugger to stop twice in this example?

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
test/mjsunit/es6/throw-caught-by-default-reject-handler.js:70:
assertEquals(["resolve q", "end main", "resolve p", "p0.then"], log);
ditto

[Yang, 2014-08-13 11:13:49]

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/reject-caug...
File test/mjsunit/es6/reject-caught-by-default-reject-handler.js (right):

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/reject-caug...
test/mjsunit/es6/reject-caught-by-default-reject-handler.js:69:
assertEquals(["resolve q", "end main", "resolve p", "p0.then"], log);
On 2014/08/13 11:03:04, aandrey wrote:
> should there be "p1.then" as well?

No. That code path is never taken, since p1 is rejected, not resolved.

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
File test/mjsunit/es6/throw-caught-by-default-reject-handler.js (right):

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
test/mjsunit/es6/throw-caught-by-default-reject-handler.js:50: // p2 is rejected
by p1's default reject handler.
On 2014/08/13 11:03:04, aandrey wrote:
> do we want the debugger to stop twice in this example?

I'm not sure. Maybe we should suppress rejections from the default reject
handler?

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
test/mjsunit/es6/throw-caught-by-default-reject-handler.js:70:
assertEquals(["resolve q", "end main", "resolve p", "p0.then"], log);
On 2014/08/13 11:03:04, aandrey wrote:
> ditto

Acknowledged.

[aandrey, 2014-08-13 11:22:03]

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
File test/mjsunit/es6/throw-caught-by-default-reject-handler.js (right):

https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
test/mjsunit/es6/throw-caught-by-default-reject-handler.js:50: // p2 is rejected
by p1's default reject handler.
On 2014/08/13 11:13:49, Yang wrote:
> On 2014/08/13 11:03:04, aandrey wrote:
> > do we want the debugger to stop twice in this example?
> 
> I'm not sure. Maybe we should suppress rejections from the default reject
> handler?

I'm not sure either, but DevTools will ignore pauses if frameCount == 0 anyway.
Not sure about error messaging though.

In the test with returning Promise.reject() I think we should give a chance for
DevTools to stop. If we have frameCount == 0 in both cases, we won't stop
obviously... Would be great to stop at return point after returning a
possibly-unhandled rejected promise.

[aandrey, 2014-08-13 11:25:42]

On 2014/08/13 11:22:03, aandrey wrote:
>
https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
> File test/mjsunit/es6/throw-caught-by-default-reject-handler.js (right):
> 
>
https://codereview.chromium.org/461023002/diff/1/test/mjsunit/es6/throw-caugh...
> test/mjsunit/es6/throw-caught-by-default-reject-handler.js:50: // p2 is
rejected
> by p1's default reject handler.
> On 2014/08/13 11:13:49, Yang wrote:
> > On 2014/08/13 11:03:04, aandrey wrote:
> > > do we want the debugger to stop twice in this example?
> > 
> > I'm not sure. Maybe we should suppress rejections from the default reject
> > handler?
> 
> I'm not sure either, but DevTools will ignore pauses if frameCount == 0
anyway.
> Not sure about error messaging though.
> 
> In the test with returning Promise.reject() I think we should give a chance
for
> DevTools to stop. If we have frameCount == 0 in both cases, we won't stop
> obviously... Would be great to stop at return point after returning a
> possibly-unhandled rejected promise.

On other thought, this behavior may be fine. We don't stop debugger on syntax
errors for example.

[Yang, 2014-08-14 06:57:56]

Committed patchset #1 manually as 23112 (presubmit successful).