Make gtk2/gtk3 compile time switchable.

Make gtk2/gtk3 compile time switchable.

This adds a "use_gtk3" gyp/gn flag, along with a separate gtk2/gtk3
targets in system.gyp and //build/config/linux/gtk[23] in gn.

BUG=132847
Committed: https://crrev.com/b8f8f682a82bce992edb35c3030214ede236d3ba
Cr-Commit-Position: refs/heads/master@{#346742}

[Elliot Glaysher, 2015-08-20 20:55:45]

erg@chromium.org changed reviewers:
+ knthzh@gmail.com, thakis@chromium.org

[Elliot Glaysher, 2015-08-20 20:55:45]

The libgtkui code doesn't compile with GTK3 in this patch. That would require
something like https://codereview.chromium.org/1234223005/ and this is to put
down a foundation for such a patch.

thakis: review

knthzh: FYI. I spent some time trying to move the current code into a loadable
shared object so that we could have swappable modules at startup. The whole
thing is actually really non-trivial and would involve more rearchitecture of
chrome than I have spare time to do right now. This patch though should actually
lay down a foundation that you can use. (In the gn build, you can switch
compiling between gtk2 and gtk3 without rebuilding all of chrome, which should
help you.)

[knthzh, 2015-08-24 20:52:01]

On 2015/08/20 20:55:45, Elliot Glaysher wrote:
> knthzh: FYI. I spent some time trying to move the current code into a loadable
> shared object so that we could have swappable modules at startup. The whole
> thing is actually really non-trivial and would involve more rearchitecture of
> chrome than I have spare time to do right now.
That's fine. Feel free to implement it later, since this and
https://codereview.chromium.org/1234223005/ ended up being surprisingly
independent. I might go for it if an edit-compile-test cycle only require
rebuilding the gtk parts.

https://codereview.chromium.org/1293073006/diff/20001/chrome/browser/ui/libgt...
File chrome/browser/ui/libgtk2ui/BUILD.gn (right):

https://codereview.chromium.org/1293073006/diff/20001/chrome/browser/ui/libgt...
chrome/browser/ui/libgtk2ui/BUILD.gn:95: "-Wno-deprecated-declarations",
This isn't necessary anymore. I found out that gtk lets me selectively silence
warnings for when use of deprecated functions is unavoidable:
https://developer.gnome.org/Deprecations/
Sorry for the trouble.

[Nico, 2015-08-25 00:23:53]

https://codereview.chromium.org/1293073006/diff/20001/build/linux/system.gyp
File build/linux/system.gyp (right):

https://codereview.chromium.org/1293073006/diff/20001/build/linux/system.gyp#...
build/linux/system.gyp:427: # Hide GTK3 and related dependencies when
use_gtk3==0 because the user
Should we hide the GTK2 packages uf use_gtk3==0?

Even better, instead of hiding them, should gtk2 and gtk3 always exist, but have
one of them be an empty type none target? (gtk2 would be empty if use_gtk3, gtk3
would be empty if !use_gtk3)

https://codereview.chromium.org/1293073006/diff/20001/chrome/browser/ui/libgt...
File chrome/browser/ui/libgtk2ui/BUILD.gn (right):

https://codereview.chromium.org/1293073006/diff/20001/chrome/browser/ui/libgt...
chrome/browser/ui/libgtk2ui/BUILD.gn:95: "-Wno-deprecated-declarations",
On 2015/08/24 20:52:01, knthzh wrote:
> This isn't necessary anymore. I found out that gtk lets me selectively silence
> warnings for when use of deprecated functions is unavoidable:
> https://developer.gnome.org/Deprecations/
> Sorry for the trouble.

Cool, remove this then I guess.

https://codereview.chromium.org/1293073006/diff/20001/gpu/gles2_conform_suppo...
File gpu/gles2_conform_support/gles2_conform_test.gyp (right):

https://codereview.chromium.org/1293073006/diff/20001/gpu/gles2_conform_suppo...
gpu/gles2_conform_support/gles2_conform_test.gyp:111: 'dependencies':
['../../build/linux/system.gyp:gtk2']
Does this intentionally depend on gtk2 unconditionally?

https://codereview.chromium.org/1293073006/diff/20001/remoting/host/BUILD.gn
File remoting/host/BUILD.gn (right):

https://codereview.chromium.org/1293073006/diff/20001/remoting/host/BUILD.gn#...
remoting/host/BUILD.gn:93: deps += [ "//build/config/linux/gtk2" ]
same question

https://codereview.chromium.org/1293073006/diff/20001/remoting/host/it2me/BUI...
File remoting/host/it2me/BUILD.gn (right):

https://codereview.chromium.org/1293073006/diff/20001/remoting/host/it2me/BUI...
remoting/host/it2me/BUILD.gn:56: deps += [ "//build/config/linux/gtk2" ]
same question

https://codereview.chromium.org/1293073006/diff/20001/remoting/remoting_host....
File remoting/remoting_host.gypi (right):

https://codereview.chromium.org/1293073006/diff/20001/remoting/remoting_host....
remoting/remoting_host.gypi:75: '../build/linux/system.gyp:gtk2',
you get the idea

https://codereview.chromium.org/1293073006/diff/20001/remoting/remoting_host....
remoting/remoting_host.gypi:677: '../build/linux/system.gyp:gtk2',
you still do

[Lambros, 2015-08-25 19:01:17]

lambroslambrou@chromium.org changed reviewers:
+ lambroslambrou@chromium.org

[Lambros, 2015-08-25 19:01:17]

https://codereview.chromium.org/1293073006/diff/20001/gpu/gles2_conform_suppo...
File gpu/gles2_conform_support/gles2_conform_test.gyp (right):

https://codereview.chromium.org/1293073006/diff/20001/gpu/gles2_conform_suppo...
gpu/gles2_conform_support/gles2_conform_test.gyp:111: 'dependencies':
['../../build/linux/system.gyp:gtk2']
On 2015/08/25 00:23:52, Nico wrote:
> Does this intentionally depend on gtk2 unconditionally?

The Chromoting host process on (non-ChromeOS) Linux uses GTK to display a couple
of dialogs. This process is a separate binary from Chrome the browser, so it
doesn't make sense for us to pull Aura into our build. So we unconditionally
depend on GTK.
As for GTK2 versus GTK3, I don't know if we need to make code changes for GTK3.
The code in question is:
https://code.google.com/p/chromium/codesearch#chromium/src/remoting/host/disc...
https://code.google.com/p/chromium/codesearch#chromium/src/remoting/host/cont...

[Elliot Glaysher, 2015-08-27 19:07:32]

https://codereview.chromium.org/1293073006/diff/20001/build/linux/system.gyp
File build/linux/system.gyp (right):

https://codereview.chromium.org/1293073006/diff/20001/build/linux/system.gyp#...
build/linux/system.gyp:427: # Hide GTK3 and related dependencies when
use_gtk3==0 because the user
On 2015/08/25 00:23:52, Nico wrote:
> Should we hide the GTK2 packages uf use_gtk3==0?
> 
> Even better, instead of hiding them, should gtk2 and gtk3 always exist, but
have
> one of them be an empty type none target? (gtk2 would be empty if use_gtk3,
gtk3
> would be empty if !use_gtk3)

I don't think so. So we still use gtk2 in a lot of test binaries / remoting /
etc, and I suspect that they won't get ported for a while. (This is my answer to
why we're unconditionally depending on gtk2 from the non-chrome targets.)

The eventual place that I want to end up is with separate dynamically loaded
gtk2 / gtk3 versions of libgtkui; so having it be an exclusive or is not what I
want. However, we need to hide the gtk3 package here because pkg-config always
runs in gyp even if we don't invoke this dependency.

[Elliot Glaysher, 2015-08-31 21:04:50]

friendly ping

[Nico, 2015-08-31 21:33:44]

lgtm

(sorry, i thought i had lgtm'd with my comments and didn't realize this was
waiting on me. protip: always say "user: ping" instead of just "ping")

[Elliot Glaysher, 2015-08-31 21:38:13]

The CQ bit was checked by erg@chromium.org

[Elliot Glaysher, 2015-08-31 21:38:48]

On 2015/08/31 21:33:44, Nico wrote:
> (sorry, i thought i had lgtm'd with my comments and didn't realize this was
> waiting on me. protip: always say "user: ping" instead of just "ping")

(sorry, i didn't realize there were other people on the review. hence your tip
for "always".)

[commit-bot: I haz the power, 2015-08-31 21:40:22]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1293073006/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1293073006/40001

[commit-bot: I haz the power, 2015-08-31 22:15:20]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-08-31 22:15:20]

Try jobs failed on following builders:
  chromium_presubmit on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)

[Elliot Glaysher, 2015-08-31 22:22:13]

erg@chromium.org changed reviewers:
+ piman@chromium.org

[Elliot Glaysher, 2015-08-31 22:22:26]

lambros, piman: owners stamps please

[Lambros, 2015-08-31 23:16:28]

remoting/ lgtm

[piman, 2015-09-01 20:33:17]

lgtm

[Elliot Glaysher, 2015-09-01 20:33:29]

The CQ bit was checked by erg@chromium.org

[commit-bot: I haz the power, 2015-09-01 20:34:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1293073006/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1293073006/40001

[commit-bot: I haz the power, 2015-09-01 21:39:09]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-09-01 21:39:57]

Patchset 3 (id:??) landed as
https://crrev.com/b8f8f682a82bce992edb35c3030214ede236d3ba
Cr-Commit-Position: refs/heads/master@{#346742}