BUG=none
TEST=build ok, run as before
Review URL: http://codereview.appspot.com/4814063

git-svn-id: https://angleproject.googlecode.com/svn/trunk@715 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@714 736b8ea6-26fd-11df-bfd4-992fa37f6226

Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@713 736b8ea6-26fd-11df-bfd4-992fa37f6226

Bug=139
Trac=#15143
Signed-off-by: Daniel Koch
Author: Michael Braithwaite

git-svn-id: https://angleproject.googlecode.com/svn/trunk@712 736b8ea6-26fd-11df-bfd4-992fa37f6226

ANGLEBUG=191
TEST=shaders with function recursion are rejected.
Review URL: http://codereview.appspot.com/4808061

git-svn-id: https://angleproject.googlecode.com/svn/trunk@711 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=188
Author: Daniel Nicoara

git-svn-id: https://angleproject.googlecode.com/svn/trunk@709 736b8ea6-26fd-11df-bfd4-992fa37f6226

Based on SVN r15167

git-svn-id: https://angleproject.googlecode.com/svn/trunk@708 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=190
Author: Makoto Kato

git-svn-id: https://angleproject.googlecode.com/svn/trunk@707 736b8ea6-26fd-11df-bfd4-992fa37f6226

Add an extension so that users can receive a device that renders in software. Currently this is done by loading a swiftshader dll.

BUG=
TEST=

Review URL: http://codereview.appspot.com/4631093

git-svn-id: https://angleproject.googlecode.com/svn/trunk@705 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@704 736b8ea6-26fd-11df-bfd4-992fa37f6226

those checked during ReadPixels. These changes are needed to avoid
having ANGLE assert in debug builds when running WebGL test cases
which provoke a GPU reset.

BUG=none
TEST=slow-shader-example and lots-of-polys-example in WebGL test suite
Review URL: http://codereview.appspot.com/4684042

git-svn-id: https://angleproject.googlecode.com/svn/trunk@703 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=104,139,179
TRAC #15143
Signed-off-by: Daniel Koch
Author: Nicolas Capens
Original-patch-by: Yore Apex

git-svn-id: https://angleproject.googlecode.com/svn/trunk@702 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=177

git-svn-id: https://angleproject.googlecode.com/svn/trunk@701 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=173
Signed-off-by: Daniel Koch
Author: Benoit Jacob

git-svn-id: https://angleproject.googlecode.com/svn/trunk@699 736b8ea6-26fd-11df-bfd4-992fa37f6226

Bug=55
TRAC #15606
Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@698 736b8ea6-26fd-11df-bfd4-992fa37f6226

Bug=55
TRAC #15606
Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@697 736b8ea6-26fd-11df-bfd4-992fa37f6226

Look at TIntermLoop::traverse() in IntermTraverse.cpp, the control init part is not handled explicitly there (unlike the other two parts).  It is expected hat visitLoop will cover the init part.  The bug in the MapLongVariableNames is that the visitLoop doesn't do anything.  This CL fixes it.

BUG=171
TEST=conformance/glsl-long-variable-names.html passing.
Review URL: http://codereview.appspot.com/4644045

git-svn-id: https://angleproject.googlecode.com/svn/trunk@696 736b8ea6-26fd-11df-bfd4-992fa37f6226

After the shader compile (and before code output), the precision scopes are already lost.  In order to correctly output precisions, we need to emit precision in each variable declaration, therefore, each variable should have its precision set.  This CL fixes the bugs that the precisions are lost for variables using default precsions and struct fields.  Also, this CL fixes a bug in the grammar: constructors are not type_specifier and they shouldn't have precisions.

BUG=168
TEST=webgl conformance tests, gles2 conformance tests.
Review URL: http://codereview.appspot.com/4617041

git-svn-id: https://angleproject.googlecode.com/svn/trunk@695 736b8ea6-26fd-11df-bfd4-992fa37f6226

BUG=none
TEST=Translator with GLES2 output backend emit precisions for function arguments and return type.
Review URL: http://codereview.appspot.com/4643042

git-svn-id: https://angleproject.googlecode.com/svn/trunk@694 736b8ea6-26fd-11df-bfd4-992fa37f6226

Some ATI cards appear not to flush the pipeline correctly, causing textures not to pick up newly-drawn contexts, so do an extra no-op draw when necessary.

BUG=169
TEST=chromium on hulu.com

Review URL: http://codereview.appspot.com/4576051

git-svn-id: https://angleproject.googlecode.com/svn/trunk@693 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/4568080

git-svn-id: https://angleproject.googlecode.com/svn/trunk@692 736b8ea6-26fd-11df-bfd4-992fa37f6226

We were setting the max vertex way too low, causing AMD cards to draw the line with an incorrect vertex.

BUG=
TEST=enable "Composited render layer borders" in chromium

Review URL: http://codereview.appspot.com/4588042

git-svn-id: https://angleproject.googlecode.com/svn/trunk@691 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@689 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@688 736b8ea6-26fd-11df-bfd4-992fa37f6226

With this CL, we have the option to select a code output backend: GLSL, GLSL ES, or HLSL.

Note that we always emit the highest supported float precision for fragment shader due to anglebug 168.  Although this is a temporary solution, it's not against GLSL ES spec, because it's ok for implementation to upgrade precision.

Tested with WebGL conformance test suite, GLES2 conformance test suite (only failed 2/1198), and a few webgl demos, including worlds of webgl, aquarium, etc.

anglebug=81
test=translator emitting correct GLSL ES code when ES2 backend is selected.
Review URL: http://codereview.appspot.com/4550129

git-svn-id: https://angleproject.googlecode.com/svn/trunk@687 736b8ea6-26fd-11df-bfd4-992fa37f6226

TBR=apatrick@chromium.org

git-svn-id: https://angleproject.googlecode.com/svn/trunk@686 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@684 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/4589043

git-svn-id: https://angleproject.googlecode.com/svn/trunk@683 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=163
TRAC #17019
Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@682 736b8ea6-26fd-11df-bfd4-992fa37f6226

Issue=160
TRAC #16817
Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@681 736b8ea6-26fd-11df-bfd4-992fa37f6226

Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@680 736b8ea6-26fd-11df-bfd4-992fa37f6226

Similar fix for cube maps to follow.
Review URL: http://codereview.appspot.com/4532124

git-svn-id: https://angleproject.googlecode.com/svn/trunk@678 736b8ea6-26fd-11df-bfd4-992fa37f6226

It turns out SetWindowLong succeeds for other threads in the same process but we do not want the WNDPROC to be called from threads other than the thread on which EGL is running on. This was one of the reasons --in-process-gpu was not working with ANGLE in Chrome.
Review URL: http://codereview.appspot.com/4576045

git-svn-id: https://angleproject.googlecode.com/svn/trunk@677 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/4559065

git-svn-id: https://angleproject.googlecode.com/svn/trunk@673 736b8ea6-26fd-11df-bfd4-992fa37f6226

The "Debug This Pixel..." feature in PIX often fails when using the
D3D9Ex interfaces.  In order to get debug pixel to work on a Vista/Win 7
machine, define "ANGLE_ENABLE_D3D9EX=0" in your project file.

Review URL: http://codereview.appspot.com/4526106

git-svn-id: https://angleproject.googlecode.com/svn/trunk@672 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/4529113

git-svn-id: https://angleproject.googlecode.com/svn/trunk@671 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/4574041

git-svn-id: https://angleproject.googlecode.com/svn/trunk@670 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@669 736b8ea6-26fd-11df-bfd4-992fa37f6226

It seems that DWM will, when given an HWND that has an associated
FlipEX swap chain, stretch the contents of the HWND to fill the new
window IF you don't issue draw calls to that HWND before the WM_PAINT
returns. This leads to horrendous sizing artifacts.

For Chromium, this is a show stopper.

But, FlipEx is definitely a good thing, for regular single process
applications.

So, is there a clean way to conditionally disable FlipEX for
these problem cases? The heuristic in this patch basically says "if your hwnd
is not in-process, then you're probably not painting in its HWND, so don't sue
FlipEX."

This catches the Chromium use case. There is still the possibility of a regular
single-process EGL application that tries to draw outside WM_PAINT that would
experience sizing artefacts. But, maybe that's OK.

Anyways, I'm not married to this technique. Just throwing something out there
to start the discussion. :)

Review URL: http://codereview.appspot.com/4536102

git-svn-id: https://angleproject.googlecode.com/svn/trunk@668 736b8ea6-26fd-11df-bfd4-992fa37f6226

Without this, sometimes the #line directive shows up at the end of
the previous line of shader source.  This was causing syntax errors
when trying to compile shaders during a PIX session.

Review URL: http://codereview.appspot.com/4561059

git-svn-id: https://angleproject.googlecode.com/svn/trunk@667 736b8ea6-26fd-11df-bfd4-992fa37f6226