Update bench_expectations.txt with new codes in issue 6842112.

Update bench_expectations.txt with new codes in issue 6842112.
Rev range is [6580, 6585].
Committed: https://code.google.com/p/skia/source/detail?r=6589

[benchen, 2012-11-28 19:58:10]

https://codereview.appspot.com/6858091 is under review, and this is result
generated from the new code.

[EricB, 2012-11-28 20:15:28]

On 2012/11/28 19:58:10, benchen wrote:
> https://codereview.appspot.com/6858091 is under review, and this is result
> generated from the new code.

Rubber stamp LGTM

[benchen, 2012-11-28 20:22:53]

Thanks for the stamp! Will submit while waiting for the other cl.
On 2012/11/28 20:15:28, EricB wrote:
> On 2012/11/28 19:58:10, benchen wrote:
> > https://codereview.appspot.com/6858091 is under review, and this is result
> > generated from the new code.
> 
> Rubber stamp LGTM

[epoger, 2012-11-29 19:55:52]

LGTM

[EricB, 2012-11-29 19:59:46]

A note on this... Looking at the log output here:
http://70.32.156.51:10115/builders/Skia_Shuttle_Ubuntu12_ATI5770_Float_Bench_...

The step *should* be failing due to bench values being out of range but instead
there's a try/catch somewhere which is making the assumption that the exception
is due to not having enough data.

[benchen, 2012-11-30 04:46:33]

This is due to Line 76 of
https://code.google.com/p/skia/source/browse/buildbot/slave/skia_slave_script...
Although I prefer the current soft fail before we confirm that we should really
make everyone panic with our alert triggering algorithm.

On 2012/11/29 19:59:46, EricB wrote:
> A note on this... Looking at the log output here:
>
http://70.32.156.51:10115/builders/Skia_Shuttle_Ubuntu12_ATI5770_Float_Bench_...
> 
> The step *should* be failing due to bench values being out of range but
instead
> there's a try/catch somewhere which is making the assumption that the
exception
> is due to not having enough data.

[EricB, 2012-11-30 12:42:20]

On 2012/11/30 04:46:33, benchen wrote:
> This is due to Line 76 of
>
https://code.google.com/p/skia/source/browse/buildbot/slave/skia_slave_script...
> Although I prefer the current soft fail before we confirm that we should
really
> make everyone panic with our alert triggering algorithm.
> 
> On 2012/11/29 19:59:46, EricB wrote:
> > A note on this... Looking at the log output here:
> >
>
http://70.32.156.51:10115/builders/Skia_Shuttle_Ubuntu12_ATI5770_Float_Bench_...
> > 
> > The step *should* be failing due to bench values being out of range but
> instead
> > there's a try/catch somewhere which is making the assumption that the
> exception
> > is due to not having enough data.

I think that's fine for now.  I just wanted to make sure we were aware of it.