InputLayoutCache only hashes input layout keys up to last used element.

InputLayoutCache only hashes input layout keys up to last used element.

R=geofflang@chromium.org

Committed: https://code.google.com/p/angleproject/source/detail?r=a612ed8

[Geoff Lang, 2013-08-08 19:48:20]

https://codereview.appspot.com/12676043/diff/1/src/libGLESv2/renderer/InputLa...
File src/libGLESv2/renderer/InputLayoutCache.cpp (right):

https://codereview.appspot.com/12676043/diff/1/src/libGLESv2/renderer/InputLa...
src/libGLESv2/renderer/InputLayoutCache.cpp:212: return std::equal(a.begin(),
a.end(), stdext::checked_array_iterator<const char*>(b.begin(),
b.elementCount));
We have some people who are building ANGLE with gcc/mingw so it may be an issue
to use the microsoft-only stdext::checked_array_iterator extension.

[apatrick1, 2013-08-12 23:02:37]

What do you think about disabling that warning? It might catch some bugs but if
we have to support mingw then it rules out much of the STL.

Also adding Shannon, didn't mean to omit from reviewers.

[shannonwoods1, 2013-08-13 20:46:38]

On 2013/08/12 23:02:37, apatrick1 wrote:
> What do you think about disabling that warning? It might catch some bugs but
if
> we have to support mingw then it rules out much of the STL.
> 
> Also adding Shannon, didn't mean to omit from reviewers.

I'm not sure how many people out there are building on mingw, but the
compiler-dependent defines for HASH_MAP at the head of the ResourceManager and
Context headers were introduced due to one contributor wanting to cross-compile:
https://codereview.appspot.com/6446083/
Apparently C++11 STL isn't a problem on that particular setup.

[apatrick1, 2013-08-15 00:02:37]

Patch set #2 should build on mingw.

[Geoff Lang, 2013-08-15 00:15:58]

On 2013/08/15 00:02:37, apatrick1 wrote:
> Patch set #2 should build on mingw.

LGTM

[apatrick1, 2013-08-15 19:43:44]

Committed patchset #3 manually as ra612ed8 (presubmit successful).