increase es5 scan timeout to 60s

chrome 30 and the r5569 changes now take more than 20s to do
an es5 scan. it's not a lot more than 20s, but I'm increasing the
timeout to 60s to give us some headroom.

[kpreid2, 2013-08-22 18:33:53]

LGTM

It would be nice to know if this is a performance regression in Chrome/our use
thereof, or if version 30 has more things for the scanner to scan.

[felix8a, 2013-08-22 19:01:08]

On 2013/08/22 18:33:53, kpreid2 wrote:
> LGTM
> 
> It would be nice to know if this is a performance regression in Chrome/our use
> thereof, or if version 30 has more things for the scanner to scan.

on the test machine,
comparing r5568 vs r5569 on Chrome 29,
"ant runtests" time before the scans is about the same (4m30s),
es53 scan goes from 60s to 100s,
es5 scan goes from 9s to 13s

comparing Chrome 29 vs Chrome 30 on r5569
"ant runtests" time before the scans is about the same (4m30s),
es53 scan goes from 100s to 120s,
es5 scan goes from 13s to 26s (timeout)

from that, seems likely that it's just that there are more things to scan,
though there's a chance that something about the scanner in particular
is less efficient in Chrome 30

[felix8a, 2013-08-22 19:02:14]

@r5572

[kpreid2, 2013-08-22 19:15:35]

On 2013/08/22 19:01:08, felix8a wrote:
> On 2013/08/22 18:33:53, kpreid2 wrote:
> > LGTM
> > 
> > It would be nice to know if this is a performance regression in Chrome/our
use
> > thereof, or if version 30 has more things for the scanner to scan.
...
> from that, seems likely that it's just that there are more things to scan,

Sorry, I should have been more specific (and I didn't actually remember we have
this info in a nice form); the scanner will print a performance summary at the
end of its console log, like so (this example is Firefox and an uncommitted
version):

    19861 references scanned. Total elapsed time 191776ms, speed 103.56/s.

Could you take a look at whether the number of references is larger?

[felix8a, 2013-08-22 19:34:47]

On 2013/08/22 19:15:35, kpreid2 wrote:
> On 2013/08/22 19:01:08, felix8a wrote:
> > On 2013/08/22 18:33:53, kpreid2 wrote:
> > > LGTM
> > > 
> > > It would be nice to know if this is a performance regression in Chrome/our
> use
> > > thereof, or if version 30 has more things for the scanner to scan.
> ...
> > from that, seems likely that it's just that there are more things to scan,
> 
> Sorry, I should have been more specific (and I didn't actually remember we
have
> this info in a nice form); the scanner will print a performance summary at the
> end of its console log, like so (this example is Firefox and an uncommitted
> version):
> 
>     19861 references scanned. Total elapsed time 191776ms, speed 103.56/s.
> 
> Could you take a look at whether the number of references is larger?

that info isn't saved right now (we're only saving fail results). I'd have to
rerun it all manually.
I'm thinking of marking some tests as "always save results" whether they fail or
not (for this and other reasons), but haven't done it yet.