Attempt at fixing reliability crash. If the AutomationProvider goes

Attempt at fixing reliability crash. If the AutomationProvider goes
away while waiting for the right notification then we could crash when
attempting to send the message.

If this is it I'll update the other observers that have a similar
problem.

BUG=63647
TEST=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=71447

[sky, 2011-01-14 00:20:10]

I'm also NULLing out reply_message_ after deleting this so that if there are
still problems we can see if we hit Observe already.

[Paweł Hajdan Jr., 2011-01-14 09:09:14]

LGTM, thank you for debugging this!

Could you also convert the VLOG in automation_provider.cc,
AutomationProvider::OnChannelError if it doesn't fire for every UI test (that
would be too much noise)? It seems it would be an important clue here.

[sky, 2011-01-14 16:29:20]

On 2011/01/14 09:09:14, Paweł Hajdan Jr. wrote:
> LGTM, thank you for debugging this!
> 
> Could you also convert the VLOG in automation_provider.cc,
> AutomationProvider::OnChannelError if it doesn't fire for every UI test (that
> would be too much noise)? It seems it would be an important clue here.

I'll do that separately so that I can land this now.

  -Scott

[sky, 2011-01-14 17:04:51]

On Fri, Jan 14, 2011 at 1:09 AM,  <phajdan.jr@chromium.org> wrote:
> LGTM, thank you for debugging this!
>
> Could you also convert the VLOG in automation_provider.cc,

Convert it LOG? What level?

  -Scott

> AutomationProvider::OnChannelError if it doesn't fire for every UI test
> (that
> would be too much noise)? It seems it would be an important clue here.
>
> http://codereview.chromium.org/6291002/
>

[Paweł Hajdan Jr., 2011-01-14 17:19:19]

I think LOG(WARNING). Thanks!

On Fri, Jan 14, 2011 at 18:04, Scott Violet <sky@chromium.org> wrote:

> On Fri, Jan 14, 2011 at 1:09 AM,  <phajdan.jr@chromium.org> wrote:
> > LGTM, thank you for debugging this!
> >
> > Could you also convert the VLOG in automation_provider.cc,
>
> Convert it LOG? What level?
>
>  -Scott
>
> > AutomationProvider::OnChannelError if it doesn't fire for every UI test
> > (that
> > would be too much noise)? It seems it would be an important clue here.
> >
> > http://codereview.chromium.org/6291002/
> >
>

[sky, 2011-01-14 18:46:51]

This results in a LOG(WARNING) for every test. See
http://build.chromium.org/p/tryserver.chromium/builders/linux/builds/8461/ste...
. I think we shouldn't bother.

  -Scott

On 2011/01/14 17:19:19, Paweł Hajdan Jr. wrote:
> I think LOG(WARNING). Thanks!
> 
> On Fri, Jan 14, 2011 at 18:04, Scott Violet <mailto:sky@chromium.org> wrote:
> 
> > On Fri, Jan 14, 2011 at 1:09 AM,  <mailto:phajdan.jr@chromium.org> wrote:
> > > LGTM, thank you for debugging this!
> > >
> > > Could you also convert the VLOG in automation_provider.cc,
> >
> > Convert it LOG? What level?
> >
> >  -Scott
> >
> > > AutomationProvider::OnChannelError if it doesn't fire for every UI test
> > > (that
> > > would be too much noise)? It seems it would be an important clue here.
> > >
> > > http://codereview.chromium.org/6291002/
> > >
> >

[Paweł Hajdan Jr., 2011-01-14 18:57:11]

On Fri, Jan 14, 2011 at 19:46, <sky@chromium.org> wrote:

> This results in a LOG(WARNING) for every test. See
>
>
http://build.chromium.org/p/tryserver.chromium/builders/linux/builds/8461/ste...
> . I think we shouldn't bother.
>

Agreed, thanks for checking.


>  -Scott
>
>
> On 2011/01/14 17:19:19, Paweł Hajdan Jr. wrote:
>
>> I think LOG(WARNING). Thanks!
>>
>
>  On Fri, Jan 14, 2011 at 18:04, Scott Violet <mailto:sky@chromium.org>
>> wrote:
>>
>
>  > On Fri, Jan 14, 2011 at 1:09 AM,  <mailto:phajdan.jr@chromium.org>
>> wrote:
>> > > LGTM, thank you for debugging this!
>> > >
>> > > Could you also convert the VLOG in automation_provider.cc,
>> >
>> > Convert it LOG? What level?
>> >
>> >  -Scott
>> >
>> > > AutomationProvider::OnChannelError if it doesn't fire for every UI
>> test
>> > > (that
>> > > would be too much noise)? It seems it would be an important clue here.
>> > >
>> > > http://codereview.chromium.org/6291002/
>> > >
>> >
>>
>
>
>
> http://codereview.chromium.org/6291002/
>