gn: Stop making |use_atk| depend on |use_gconf|.

gn: Stop making |use_atk| depend on |use_gconf|.

This partly reverts commit 2a395c69f ("make use of existing gn args in
ui build config").

atk itself does not depend on gconf, and the change introduced in the
commit above was likely created before commit 95ba4446 ("Move linux
pkg_config() calls into separate BUILD.gn files"), when atk and gconf
were part of the same if block in build/config/linux/BUILD.gn.

See also: https://codereview.chromium.org/2241513002

R=brettw@chromium.org,dpranke@chromium.org,mostynb@opera.com
BUG=632297
Committed: https://crrev.com/984335951bbc9b2d9532db8b8d1161494f79a105
Cr-Commit-Position: refs/heads/master@{#412634}

[Mostyn Bramley-Moore, 2016-08-16 10:15:36]

Some background from 2a395c69f...

//ui/accessibility previously always depended on gconf and glib if atk was
enabled.  IIRC this caused pkg-config errors when building without gconf and
glib installed, even if we weren't planning to build the //ui/accessibility
target.  Moving the gconf and glib deps into conditional blocks skipped these
unnecessary pkg-config calls.

On a sample ubuntu (14.04) machine, `pkg-config --libs atk` says "-latk-1.0
-lgobject-2.0 -lglib-2.0", so I have a feeling that it really does depend on
glib at least.

[Raphael Kubo da Costa (rakuco), 2016-08-16 10:19:17]

atk does depend on glib, but not gconf. In my case it's OK to depend on the
former, but depending on the latter pulls in additional dependencies that are
really not needed.

If this patch causes trouble for you, instead of just removing the |use_gconf|
bit I can replace it with |use_glib|.

[Mostyn Bramley-Moore, 2016-08-16 11:02:01]

On 2016/08/16 10:19:17, Raphael Kubo da Costa (rakuco) wrote:
> atk does depend on glib, but not gconf. In my case it's OK to depend on the
> former, but depending on the latter pulls in additional dependencies that are
> really not needed.
> 
> If this patch causes trouble for you, instead of just removing the |use_gconf|
> bit I can replace it with |use_glib|.

This change shouldn't cause trouble for me.

I'm not sure if we should replace the use_gconf with use_glib here.  An
alternative would be to add an assert(use_glib) inside a conditional use_atk
block somewhere, or just let the use_atk=true,use_glib=false configuration break
at ninja-time.

[mforce2, 2016-08-16 18:57:33]

On 2016/08/16 11:02:01, Mostyn Bramley-Moore wrote:
> On 2016/08/16 10:19:17, Raphael Kubo da Costa (rakuco) wrote:
> > atk does depend on glib, but not gconf. In my case it's OK to depend on the
> > former, but depending on the latter pulls in additional dependencies that
are
> > really not needed.
> > 
> > If this patch causes trouble for you, instead of just removing the
|use_gconf|
> > bit I can replace it with |use_glib|.
> 
> This change shouldn't cause trouble for me.
> 
> I'm not sure if we should replace the use_gconf with use_glib here.  An
> alternative would be to add an assert(use_glib) inside a conditional use_atk
> block somewhere, or just let the use_atk=true,use_glib=false configuration
break
> at ninja-time.

 In my opinion as a distribution package maintainer it would be better to have
the assert and the warning rather than the configuration break at ninja-time. It
saves some time when you get errors as clear and as early as possible.

[brettw, 2016-08-16 21:18:57]

I really like not linking the .gni files, so removing the import here sounds
great.

Is there a good build file where we can put the assert? I'm thinking maybe next
to whatever "atk" is.

[Mostyn Bramley-Moore, 2016-08-17 00:44:43]

> Is there a good build file where we can put the assert? I'm thinking maybe
next
> to whatever "atk" is.

How about this? https://codereview.chromium.org/2251673004/

[brettw, 2016-08-17 18:06:32]

LGTM for this part, that other patch should address the issue.

[Raphael Kubo da Costa (rakuco), 2016-08-17 20:10:14]

The CQ bit was checked by raphael.kubo.da.costa@intel.com

[commit-bot: I haz the power, 2016-08-17 20:10:44]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-08-17 20:59:59]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-08-17 21:05:38]

Description was changed from

==========
gn: Stop making |use_atk| depend on |use_gconf|.

This partly reverts commit 2a395c69f ("make use of existing gn args in
ui build config").

atk itself does not depend on gconf, and the change introduced in the
commit above was likely created before commit 95ba4446 ("Move linux
pkg_config() calls into separate BUILD.gn files"), when atk and gconf
were part of the same if block in build/config/linux/BUILD.gn.

See also: https://codereview.chromium.org/2241513002

R=brettw@chromium.org,dpranke@chromium.org,mostynb@opera.com
BUG=632297
==========

to

==========
gn: Stop making |use_atk| depend on |use_gconf|.

This partly reverts commit 2a395c69f ("make use of existing gn args in
ui build config").

atk itself does not depend on gconf, and the change introduced in the
commit above was likely created before commit 95ba4446 ("Move linux
pkg_config() calls into separate BUILD.gn files"), when atk and gconf
were part of the same if block in build/config/linux/BUILD.gn.

See also: https://codereview.chromium.org/2241513002

R=brettw@chromium.org,dpranke@chromium.org,mostynb@opera.com
BUG=632297

Committed: https://crrev.com/984335951bbc9b2d9532db8b8d1161494f79a105
Cr-Commit-Position: refs/heads/master@{#412634}
==========

[commit-bot: I haz the power, 2016-08-17 21:05:39]

Patchset 1 (id:??) landed as
https://crrev.com/984335951bbc9b2d9532db8b8d1161494f79a105
Cr-Commit-Position: refs/heads/master@{#412634}