gpu/khronos_glcts_support: add support for GN builds

gpu/khronos_glcts_support: add support for GN builds

Port the GYP configs to BUILD.gn so that this test
can be built with GN.

BUG=chromium:412865, chromium:471903
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org, brettw@chromium.org
TEST=Verify khronos_glcts_test builds and runs with GN build

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>;

gpu/khronos_glcts_support: Move target_defaults into separate file

Move the target_defaults into its own gypi file so that we can
parse the variables in khronos_glcts.gypi with the gypi_to_gn.py
script.  The gypi_to_gn.py chokes on target_defaults.

BUG=chromium:412865, chromium:471903
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org
TEST=Verify khronos_glcts_test builds and runs

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>;

gpu/khronos_glcts_support: use override specifier

Annotate the base class override functions with the
override specifier to appease the chromium-style
checker.

BUG=chromium:412865
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org
TEST=Verify khronos_glcts_test builds and runs

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>;

[piman, 2015-10-28 22:45:38]

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
File gpu/khronos_glcts_support/BUILD.gn (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
gpu/khronos_glcts_support/BUILD.gn:75: configs += [
"//build/config/compiler:rtti" ]
I see this duplicated many times in all the targets. Is there a way to share?

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/k...
File gpu/khronos_glcts_support/khronos_glcts_defaults.gypi (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/k...
gpu/khronos_glcts_support/khronos_glcts_defaults.gypi:25:
['_type=="static_library"', {
Why specific to static_library?

[Dirk Pranke, 2015-10-28 22:55:35]

The GN changes mostly lgtm stylistically at least. I defer to others to ensure
their correctness.

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
File gpu/khronos_glcts_support/BUILD.gn (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
gpu/khronos_glcts_support/BUILD.gn:75: configs += [
"//build/config/compiler:rtti" ]
On 2015/10/28 22:45:38, piman (slow to review) wrote:
> I see this duplicated many times in all the targets. Is there a way to share?

You could probably create a template that wrapped source_set() and did 
the appropriate adjustments, but I'm not sure if it would be an overall 
win. It might be worth trying.

I'm not sure that there are any other easy alternatives.

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
gpu/khronos_glcts_support/BUILD.gn:87: }
nit: blank lines between top-level things?

[U. Artie Eoff, 2015-10-28 22:57:31]

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
File gpu/khronos_glcts_support/BUILD.gn (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
gpu/khronos_glcts_support/BUILD.gn:75: configs += [
"//build/config/compiler:rtti" ]
On 2015/10/28 22:45:38, piman (slow to review) wrote:
> I see this duplicated many times in all the targets. Is there a way to share?

I could not figure out how to share it in GN

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/k...
File gpu/khronos_glcts_support/khronos_glcts_defaults.gypi (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/k...
gpu/khronos_glcts_support/khronos_glcts_defaults.gypi:25:
['_type=="static_library"', {
On 2015/10/28 22:45:38, piman (slow to review) wrote:
> Why specific to static_library? 

Because all the khronos_glcts third_party code is compiled into static_libraries
and needs rtti to compile (due to use of dynamic_cast).  However, once we link
everything into any of the final executable(s), we need -fno-rtti, otherwise
native/main.cc fails to link due to "undefined reference to 'typeinfo for
base::MessageLoop'" since "base" was compiled with no-rtti.

Note that this section isn't new code... it was only cut and pasted from
khronos_glcts.gypi.

[U. Artie Eoff, 2015-10-28 23:00:00]

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
File gpu/khronos_glcts_support/BUILD.gn (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
gpu/khronos_glcts_support/BUILD.gn:75: configs += [
"//build/config/compiler:rtti" ]
On 2015/10/28 22:55:35, Dirk Pranke wrote:
> On 2015/10/28 22:45:38, piman (slow to review) wrote:
> > I see this duplicated many times in all the targets. Is there a way to
share?
> 
> You could probably create a template that wrapped source_set() and did 
> the appropriate adjustments, but I'm not sure if it would be an overall 
> win. It might be worth trying.
> 
> I'm not sure that there are any other easy alternatives.

I could not figure out how to get templates to work for this... I tried.

[Dirk Pranke, 2015-10-28 23:05:30]

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
File gpu/khronos_glcts_support/BUILD.gn (right):

https://codereview.chromium.org/1405133010/diff/1/gpu/khronos_glcts_support/B...
gpu/khronos_glcts_support/BUILD.gn:75: configs += [
"//build/config/compiler:rtti" ]
On 2015/10/28 22:59:59, U. Artie Eoff wrote:
> On 2015/10/28 22:55:35, Dirk Pranke wrote:
> > On 2015/10/28 22:45:38, piman (slow to review) wrote:
> > > I see this duplicated many times in all the targets. Is there a way to
> share?
> > 
> > You could probably create a template that wrapped source_set() and did 
> > the appropriate adjustments, but I'm not sure if it would be an overall 
> > win. It might be worth trying.
> > 
> > I'm not sure that there are any other easy alternatives.
> 
> I could not figure out how to get templates to work for this... I tried.

Hmm. I think you should be able to get something to work.

Maybe land this as-is and we can see if we can get something working in a
follow-up CL?

[U. Artie Eoff, 2015-10-28 23:06:23]

Quick question... I created these as three separate commits in git.  However,
they uploaded here (git cl upload) as one large commit.  Will these be committed
as three commits or one?

[Dirk Pranke, 2015-10-28 23:07:16]

On 2015/10/28 23:06:23, U. Artie Eoff wrote:
> Quick question... I created these as three separate commits in git.  However,
> they uploaded here (git cl upload) as one large commit.  Will these be
committed
> as three commits or one?

A single commit when landed via the CQ or `git cl land`.

[U. Artie Eoff, 2015-10-28 23:09:06]

On 2015/10/28 23:07:16, Dirk Pranke wrote:
> On 2015/10/28 23:06:23, U. Artie Eoff wrote:
> > Quick question... I created these as three separate commits in git. 
However,
> > they uploaded here (git cl upload) as one large commit.  Will these be
> committed
> > as three commits or one?
> 
> A single commit when landed via the CQ or `git cl land`.

eww.. how do we prevent that and get 3 separate commits instead?

[Dirk Pranke, 2015-10-28 23:11:36]

On 2015/10/28 23:09:06, U. Artie Eoff wrote:
> eww.. how do we prevent that and get 3 separate commits instead?

You would have to split this up into 3 different CLs.

[U. Artie Eoff, 2015-10-28 23:17:28]

On 2015/10/28 23:11:36, Dirk Pranke wrote:
> On 2015/10/28 23:09:06, U. Artie Eoff wrote:
> > eww.. how do we prevent that and get 3 separate commits instead?
> 
> You would have to split this up into 3 different CLs.

Ok, I'd rather do that.  How do I abandon this?  And will Rietveld be smart
enough if I do 'git cl upload' for each commit on top of the other (e.g.
CQ-DEPEND).

[Dirk Pranke, 2015-10-28 23:22:55]

On 2015/10/28 23:17:28, U. Artie Eoff wrote:
> On 2015/10/28 23:11:36, Dirk Pranke wrote:
> > On 2015/10/28 23:09:06, U. Artie Eoff wrote:
> > > eww.. how do we prevent that and get 3 separate commits instead?
> > 
> > You would have to split this up into 3 different CLs.
> 
> Ok, I'd rather do that.  How do I abandon this?

Just close it via the 'edit issue' link.

>  And will Rietveld be smart
> enough if I do 'git cl upload' for each commit on top of the other (e.g.
> CQ-DEPEND).

Yes.

[Dirk Pranke, 2015-10-28 23:23:34]

On 2015/10/28 23:22:55, Dirk Pranke wrote:
> On 2015/10/28 23:17:28, U. Artie Eoff wrote:
> > On 2015/10/28 23:11:36, Dirk Pranke wrote:
> > > On 2015/10/28 23:09:06, U. Artie Eoff wrote:
> > > > eww.. how do we prevent that and get 3 separate commits instead?
> > > 
> > > You would have to split this up into 3 different CLs.
> > 
> > Ok, I'd rather do that.  How do I abandon this?
> 
> Just close it via the 'edit issue' link.
> 
> >  And will Rietveld be smart
> > enough if I do 'git cl upload' for each commit on top of the other (e.g.
> > CQ-DEPEND).
> 
> Yes.

To be clear: you will need to create 3 separate branches in git, with each the
parent of the next. git-cl operates per-branch, not per-commit.

[U. Artie Eoff, 2015-10-28 23:45:40]

Description was changed from

==========
gpu/khronos_glcts_support: add support for GN builds

Port the GYP configs to BUILD.gn so that this test
can be built with GN.

BUG=chromium:412865, chromium:471903
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org,
brettw@chromium.org
TEST=Verify khronos_glcts_test builds and runs with GN build

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

gpu/khronos_glcts_support: Move target_defaults into separate file

Move the target_defaults into its own gypi file so that we can
parse the variables in khronos_glcts.gypi with the gypi_to_gn.py
script.  The gypi_to_gn.py chokes on target_defaults.

BUG=chromium:412865, chromium:471903
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org
TEST=Verify khronos_glcts_test builds and runs

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

gpu/khronos_glcts_support: use override specifier

Annotate the base class override functions with the
override specifier to appease the chromium-style
checker.

BUG=chromium:412865
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org
TEST=Verify khronos_glcts_test builds and runs

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
==========

to

==========
gpu/khronos_glcts_support: add support for GN builds

Port the GYP configs to BUILD.gn so that this test
can be built with GN.

BUG=chromium:412865, chromium:471903
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org,
brettw@chromium.org
TEST=Verify khronos_glcts_test builds and runs with GN build

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

gpu/khronos_glcts_support: Move target_defaults into separate file

Move the target_defaults into its own gypi file so that we can
parse the variables in khronos_glcts.gypi with the gypi_to_gn.py
script.  The gypi_to_gn.py chokes on target_defaults.

BUG=chromium:412865, chromium:471903
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org
TEST=Verify khronos_glcts_test builds and runs

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

gpu/khronos_glcts_support: use override specifier

Annotate the base class override functions with the
override specifier to appease the chromium-style
checker.

BUG=chromium:412865
R=kbr@chromium.org, piman@chromium.org, dpranke@chromium.org
TEST=Verify khronos_glcts_test builds and runs

Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
==========