DevTools: autoresume execution on parse errors.

DevTools: autoresume execution on parse errors.

BUG=22852
TEST=DevToolsSanityTest.TestAutoContinueOnSyntaxError

Commited http://src.chromium.org/viewvc/chrome?view=rev&revision=27067

[pfeldman, 2009-09-24 14:54:50]

LG overall, some comments inside.

http://codereview.chromium.org/218026/diff/16/1003
File webkit/glue/devtools/js/tests.js (right):

http://codereview.chromium.org/218026/diff/16/1003#newcode683
Line 683: test.assertEquals(1, options.length,
this will need to be changed once we remove api from the list of scripts. why
not to test that there is _no_ unparsed script instead?

http://codereview.chromium.org/218026/diff/16/1003#newcode705
Line 705: 'handleExceptionEvent_',
i'd suggest that this is tested against WebKit api as well.

[yurys, 2009-09-24 15:12:08]

http://codereview.chromium.org/218026/diff/16/1003
File webkit/glue/devtools/js/tests.js (right):

http://codereview.chromium.org/218026/diff/16/1003#newcode683
Line 683: test.assertEquals(1, options.length,
On 2009/09/24 14:54:50, pfeldman wrote:
> this will need to be changed once we remove api from the list of scripts. why
> not to test that there is _no_ unparsed script instead?

Done.

http://codereview.chromium.org/218026/diff/16/1003#newcode705
Line 705: 'handleExceptionEvent_',
On 2009/09/24 14:54:50, pfeldman wrote:
> i'd suggest that this is tested against WebKit api as well.

I'd be glad to test it that way but I don't see how at the moment.

[Søren Thygesen Gjesse, 2009-09-25 06:28:45]

LGTM, but consider this

http://codereview.chromium.org/218026/diff/1011/18
File webkit/glue/devtools/js/debugger_agent.js (right):

http://codereview.chromium.org/218026/diff/1011/18#newcode795
Line 795: if (this.pauseOnExceptions_ && body.script) {
The assumption that if there is no script information then it is a parsing error
seems fragile. Changes to V8 might break that. I think we should aim for adding
information to the exception event which will make this check more robust.

[yurys, 2009-09-25 14:01:55]

http://codereview.chromium.org/218026/diff/1011/18
File webkit/glue/devtools/js/debugger_agent.js (right):

http://codereview.chromium.org/218026/diff/1011/18#newcode795
Line 795: if (this.pauseOnExceptions_ && body.script) {
On 2009/09/25 06:28:46, Søren Gjesse wrote:
> The assumption that if there is no script information then it is a parsing
error
> seems fragile. Changes to V8 might break that. I think we should aim for
adding
> information to the exception event which will make this check more robust.

Are there other cases when body.script is undefined except parsing errors or we
can just add something like body.isParsingError flag?

[Søren Thygesen Gjesse, 2009-09-28 07:40:10]

On 2009/09/25 14:01:55, Yury Semikhatsky wrote:
> http://codereview.chromium.org/218026/diff/1011/18
> File webkit/glue/devtools/js/debugger_agent.js (right):
> 
> http://codereview.chromium.org/218026/diff/1011/18#newcode795
> Line 795: if (this.pauseOnExceptions_ && body.script) {
> On 2009/09/25 06:28:46, Søren Gjesse wrote:
> > The assumption that if there is no script information then it is a parsing
> error
> > seems fragile. Changes to V8 might break that. I think we should aim for
> adding
> > information to the exception event which will make this check more robust.
> 
> Are there other cases when body.script is undefined except parsing errors or
we
> can just add something like body.isParsingError flag?

I think we should stick to using the error types defined in the ECMA-262 error
types (EvalError, RangeError, ReferenceError, SyntaxError, TypeError and
URIError). The ErrorMirror have the constructor function, which will be
SyntaxError in this case.