Support precision for sampler types.

Support precision for sampler types.

BUG=
R=alokp@chromium.org

Committed: https://code.google.com/p/angleproject/source/detail?r=9754ff4

[Zhenyao Mo, 2013-08-16 19:27:22]

Alok: please review.

Shannon, Nicolas, Ken, Al, FYI.

Now that we enforce uniform precision match between vx/fs shaders, it's time to
fix this long standing bug.  Otherwise a few ASSERT will fail.

Tested with webgl conformance tests and a bunch of other webgl demos, no
regression.

[Alok Priyadarshi, 2013-08-19 16:36:24]

https://codereview.appspot.com/12833045/diff/3001/src/compiler/Compiler.cpp
File src/compiler/Compiler.cpp (right):

https://codereview.appspot.com/12833045/diff/3001/src/compiler/Compiler.cpp#n...
src/compiler/Compiler.cpp:254: for (int samplerType = EbtGuardSamplerBegin + 1;
Unless I am missing something, the code blocks for both fragment and vertex
shader are the same. Can you duplicate code??

https://codereview.appspot.com/12833045/diff/3001/src/compiler/SymbolTable.h
File src/compiler/SymbolTable.h (right):

https://codereview.appspot.com/12833045/diff/3001/src/compiler/SymbolTable.h#...
src/compiler/SymbolTable.h:340: if (type.type != EbtFloat && type.type != EbtInt
&& !IsSampler(type.type))
nit: I think it would be nice to have a function like:
bool supportsPrecision(TBasicType)

and use it in both set/getDefaultPrecision.

[Zhenyao Mo, 2013-08-19 17:47:19]

please take another look.

https://codereview.appspot.com/12833045/diff/3001/src/compiler/Compiler.cpp
File src/compiler/Compiler.cpp (right):

https://codereview.appspot.com/12833045/diff/3001/src/compiler/Compiler.cpp#n...
src/compiler/Compiler.cpp:254: for (int samplerType = EbtGuardSamplerBegin + 1;
On 2013/08/19 16:36:24, Alok Priyadarshi wrote:
> Unless I am missing something, the code blocks for both fragment and vertex
> shader are the same. Can you duplicate code??

Done.

[Alok Priyadarshi, 2013-08-19 17:49:33]

lgtm with a nit

https://codereview.appspot.com/12833045/diff/10001/src/compiler/SymbolTable.h
File src/compiler/SymbolTable.h (right):

https://codereview.appspot.com/12833045/diff/10001/src/compiler/SymbolTable.h...
src/compiler/SymbolTable.h:341: return false; // Only set default precision for
int/float/sampler
nit: the comment should move to supportsPrecision()

[Zhenyao Mo, 2013-08-19 17:53:42]

https://codereview.appspot.com/12833045/diff/10001/src/compiler/SymbolTable.h
File src/compiler/SymbolTable.h (right):

https://codereview.appspot.com/12833045/diff/10001/src/compiler/SymbolTable.h...
src/compiler/SymbolTable.h:341: return false; // Only set default precision for
int/float/sampler
On 2013/08/19 17:49:33, Alok Priyadarshi wrote:
> nit: the comment should move to supportsPrecision()

Done.

[Zhenyao Mo, 2013-08-19 17:55:03]

Committed patchset #4 manually as r9754ff4 (presubmit successful).