cc: Guard overdraw metrics behind the --trace-overdraw command-line flag.

cc: Guard overdraw metrics behind the --trace-overdraw command-line flag.

Because these metrics are slow, they change the characteristics of traces. When
we want to look at performance, then these become misleading. So, enable them
only when desired, when someone wants tracing information about overdraw.

Also reverting changes made to make the overdraw metrics faster and less precise.
Since they are optional, we should get the most accurate data out of them when
they are turned on.

BUG=159718
R=enne,jamesr

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=167228

[jamesr, 2012-11-12 17:44:43]

what does using cc::Settings buy us here over just checking the command line?

I think you need to update
RenderProcessHostImpl::PropagateBrowserCommandLineToRenderer() if you want this
to work in the renderer process

[danakj, 2012-11-12 17:45:59]

On 2012/11/12 17:44:43, jamesr wrote:
> what does using cc::Settings buy us here over just checking the command line?

Potentially faster since we only parse text once? We don't check the command
line directly anywhere except settings.cc at the moment, so I'm mostly following
suit. Perhaps we should set a guideline for how we'd like this done in future.
What do you prefer?

> 
> I think you need to update
> RenderProcessHostImpl::PropagateBrowserCommandLineToRenderer() if you want
this
> to work in the renderer process

Ah! okay thanks.

[enne (OOO), 2012-11-12 17:59:39]

My guideline for the future is that we should destroy cc::Settings with fire and
put everything in LayerTreeSettings.

[danakj, 2012-11-12 18:01:02]

On 2012/11/12 17:45:59, danakj wrote:
> > I think you need to update
> > RenderProcessHostImpl::PropagateBrowserCommandLineToRenderer() if you want
> this
> > to work in the renderer process
> 
> Ah! okay thanks.

Done.

[jamesr, 2012-11-12 18:04:53]

lgtm

cc::Settings should DIAF, but we can do that separately IMO. This is a big perf
win

[commit-bot: I haz the power, 2012-11-12 18:08:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/danakj@chromium.org/11369188/2005

[commit-bot: I haz the power, 2012-11-12 18:08:20]

Presubmit check for 11369188-2005 failed and returned exit status 1.


Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for files in these directories:
    content/browser/renderer_host

Presubmit checks took 1.1s to calculate.

Was the presubmit check useful? Please send feedback & hate mail to
maruel@chromium.org!

[danakj, 2012-11-12 18:12:20]

+sky for content/browser/

[Ian Vollick, 2012-11-12 18:20:56]

On 2012/11/12 17:59:39, enne wrote:
> My guideline for the future is that we should destroy cc::Settings with fire
and
> put everything in LayerTreeSettings.

+1 -- It was only done so that we could add animations without the
LayerTreeHost, but that turned out to cause correctness issues, so there's no
point for the statics -- they should absolutely go away.

[jam, 2012-11-12 19:03:37]

lgtm

[commit-bot: I haz the power, 2012-11-12 19:04:03]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/danakj@chromium.org/11369188/2005