search registered rtconf variables backwards and break when we find the

search registered rtconf variables backwards and break when we find the 
first match.  Patch from mozilla

BUG=
R=mtklein@google.com

Committed: https://code.google.com/p/skia/source/detail?r=10337

[mtklein, 2013-07-24 21:36:12]

Am I reading this right?  Either way, last one wins?  Is this purely a speed
thing, or does it fix a bug?

[humper, 2013-07-24 21:39:47]

On 2013/07/24 21:36:12, mtklein wrote:
> Am I reading this right?  Either way, last one wins?  Is this purely a speed
> thing, or does it fix a bug?

Just a speed optimization; search changed to be backwards to be semantically
equivalent to the previous code, and no need to keep looking once we've found
the last one.

In reality there should only ever be one, so I think it's irrelevant which way
we search.

[mtklein, 2013-07-24 21:42:14]

On 2013/07/24 21:39:47, humper wrote:
> On 2013/07/24 21:36:12, mtklein wrote:
> > Am I reading this right?  Either way, last one wins?  Is this purely a speed
> > thing, or does it fix a bug?
> 
> Just a speed optimization; search changed to be backwards to be semantically
> equivalent to the previous code, and no need to keep looking once we've found
> the last one.
> 
> In reality there should only ever be one, so I think it's irrelevant which way
> we search.

Gotcha.  lgtm

I hate to be mean and call this pointless, but was there a backstory that
motivated this?  Is rtconf in the critical path anywhere?

[humper, 2013-07-24 21:44:14]

On 2013/07/24 21:42:14, mtklein wrote:
> On 2013/07/24 21:39:47, humper wrote:
> > On 2013/07/24 21:36:12, mtklein wrote:
> > > Am I reading this right?  Either way, last one wins?  Is this purely a
speed
> > > thing, or does it fix a bug?
> > 
> > Just a speed optimization; search changed to be backwards to be semantically
> > equivalent to the previous code, and no need to keep looking once we've
found
> > the last one.
> > 
> > In reality there should only ever be one, so I think it's irrelevant which
way
> > we search.
> 
> Gotcha.  lgtm
> 
> I hate to be mean and call this pointless, but was there a backstory that
> motivated this?  Is rtconf in the critical path anywhere?

Rtconf being on the critical path will always be the dream.

Patch was submitted to bugzilla, brought to my attention, I agreed that the new
code was better than the old code, and that was good enough for me :)

[humper, 2013-07-24 21:44:51]

Committed patchset #1 manually as r10337 (presubmit successful).