Update blimp_engine.gn to use headless ozone platform

Update blimp_engine.gn to use headless ozone platform

Disable ozone auto platform selection for blimp engine builds.  Ensure
blimp does not have a dependency on X11.

BUG=602312
Committed: https://crrev.com/83282ff073fe045d06df62468e9683db17f18e8c
Cr-Commit-Position: refs/heads/master@{#386469}

[maniscalco, 2016-04-11 17:34:24]

maniscalco@chromium.org changed reviewers:
+ dpranke@chromium.org, kylechar@chromium.org

[maniscalco, 2016-04-11 17:34:24]

Dirk, Kyle, would you two please review this change?

https://crrev.com/8d0dcabc234d92f418bcf1524b481293d771d03a ended up breaking
blimp by adding a dependency on X11.  Previously we had been using "use_ozone =
true" to avoid picking up an X11 dep on linux.  What's the right way to do this?
 Does this change use the ozone flags correctly?

[kylechar, 2016-04-11 17:43:16]

On 2016/04/11 at 17:34:24, maniscalco wrote:
> Dirk, Kyle, would you two please review this change?
> 
> https://crrev.com/8d0dcabc234d92f418bcf1524b481293d771d03a ended up breaking
blimp by adding a dependency on X11.  Previously we had been using "use_ozone =
true" to avoid picking up an X11 dep on linux.  What's the right way to do this?
 Does this change use the ozone flags correctly?

Oh sorry about the breakage. That is the correct way to specify a specific ozone
platform and headless is most complete platform aside from gbm/x11.

[Dirk Pranke, 2016-04-11 17:44:04]

The linked change concerns me (I wasn't aware of it until now), and I'm not sure
it was the right thing 
to do.

I feel like we probably have assumptions scattered throughout the build (not
just in blimp) that assume
that if use_ozone=true there are no dependencies on X.

Let's see what Kyle says.

[maniscalco, 2016-04-11 18:21:45]

Dirk, agreed on assumptions about use_ozone=true --> no X.

Kyle, no worries about the breakage.  I must admit, I'm not super familiar with
ozone.  Is ozone synonymous with "no X"?  Is that a bad assumption?


On 2016/04/11 17:44:04, Dirk Pranke wrote:
> The linked change concerns me (I wasn't aware of it until now), and I'm not
sure
> it was the right thing 
> to do.
> 
> I feel like we probably have assumptions scattered throughout the build (not
> just in blimp) that assume
> that if use_ozone=true there are no dependencies on X.
> 
> Let's see what Kyle says.

[kylechar, 2016-04-11 18:34:31]

The ozone framework is platform independent, so use_ozone=true itself doesn't
imply a dependency on x11. Now that there is a X11 ozone platform we want it to
be compiled by default for development and testing, hence adding it if
ozone_auto_platforms = true. If this is problematic the change can be rolled
back and something could be done.

It's probably best to specify the ozone platform you want regardless though?

On 2016/04/11 at 18:21:45, maniscalco wrote:
> Dirk, agreed on assumptions about use_ozone=true --> no X.
> 
> Kyle, no worries about the breakage.  I must admit, I'm not super familiar
with ozone.  Is ozone synonymous with "no X"?  Is that a bad assumption?
> 
> 
> On 2016/04/11 17:44:04, Dirk Pranke wrote:
> > The linked change concerns me (I wasn't aware of it until now), and I'm not
sure
> > it was the right thing 
> > to do.
> > 
> > I feel like we probably have assumptions scattered throughout the build (not
> > just in blimp) that assume
> > that if use_ozone=true there are no dependencies on X.
> > 
> > Let's see what Kyle says.

[Dirk Pranke, 2016-04-11 18:47:39]

On 2016/04/11 18:34:31, kylechar wrote:
> The ozone framework is platform independent, so use_ozone=true itself doesn't
> imply a dependency on x11. Now that there is a X11 ozone platform we want it
to
> be compiled by default for development and testing, hence adding it if
> ozone_auto_platforms = true. If this is problematic the change can be rolled
> back and something could be done.
> 
> It's probably best to specify the ozone platform you want regardless though?

I'm not sure how many people outside of those working on ozone implementations
are aware that there's more than one of them.

Perhaps it's best for someone on the ozone team to audit uses of use_ozone?

In the specific case of blimp I agree it's probably best for them to specify
headless explicitly and explicitly turn off ozone_platform_x11.

We should probably also make sure blimp gets trybots and continuous bots
to catch these sorts of things faster, and also have blimp use GN's
assert_no_deps()
to make sure we end up growing dependencies like this.

-- Dirk

[kylechar, 2016-04-11 19:18:17]

lgtm

[maniscalco, 2016-04-11 19:36:07]

Kyle, thanks for the explanation.

Dirk, good idea on using assert_no_deps().  I'll update this patch to do that. 
I'm assuming //build/config/linux:x11 would be the target I want to
assert_no_deps on.


On 2016/04/11 19:18:17, kylechar wrote:
> lgtm

[maniscalco, 2016-04-11 19:42:04]

Hmm... actually scratch that.  //build/config/linux:x11 probably isn't the right
thing to assert_no_deps on.  Either way, I'll do the assert_no_deps thing in a
follow up patch.

On 2016/04/11 19:36:07, maniscalco wrote:
> Kyle, thanks for the explanation.
> 
> Dirk, good idea on using assert_no_deps().  I'll update this patch to do that.

> I'm assuming //build/config/linux:x11 would be the target I want to
> assert_no_deps on.
> 
> 
> On 2016/04/11 19:18:17, kylechar wrote:
> > lgtm

[maniscalco, 2016-04-11 20:38:25]

Dirk, PTAL when you get a chance.  I'm eager to get this change to
blimp_engine.gn landed so our next buildbot cycle will produce a working build.

On 2016/04/11 19:42:04, maniscalco wrote:
> Hmm... actually scratch that.  //build/config/linux:x11 probably isn't the
right
> thing to assert_no_deps on.  Either way, I'll do the assert_no_deps thing in a
> follow up patch.
> 
> On 2016/04/11 19:36:07, maniscalco wrote:
> > Kyle, thanks for the explanation.
> > 
> > Dirk, good idea on using assert_no_deps().  I'll update this patch to do
that.
> 
> > I'm assuming //build/config/linux:x11 would be the target I want to
> > assert_no_deps on.
> > 
> > 
> > On 2016/04/11 19:18:17, kylechar wrote:
> > > lgtm

[Dirk Pranke, 2016-04-11 20:55:27]

lgtm

[maniscalco, 2016-04-11 20:56:08]

The CQ bit was checked by maniscalco@chromium.org

[commit-bot: I haz the power, 2016-04-11 20:57:13]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1881493003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1881493003/1

[commit-bot: I haz the power, 2016-04-11 21:03:50]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-04-11 21:05:50]

Description was changed from

==========
Update blimp_engine.gn to use headless ozone platform

Disable ozone auto platform selection for blimp engine builds.  Ensure
blimp does not have a dependency on X11.

BUG=602312
==========

to

==========
Update blimp_engine.gn to use headless ozone platform

Disable ozone auto platform selection for blimp engine builds.  Ensure
blimp does not have a dependency on X11.

BUG=602312

Committed: https://crrev.com/83282ff073fe045d06df62468e9683db17f18e8c
Cr-Commit-Position: refs/heads/master@{#386469}
==========

[commit-bot: I haz the power, 2016-04-11 21:05:51]

Patchset 1 (id:??) landed as
https://crrev.com/83282ff073fe045d06df62468e9683db17f18e8c
Cr-Commit-Position: refs/heads/master@{#386469}