Relanding this as the previous attempt failed due to build errors.

Relanding this as the previous attempt failed due to build errors.

This is continuation of the ongoing work for bug https://code.google.com/p/chromium/issues/detail?can=2&q=237249
which is splitting chrome.dll into a browser and renderer component due to build issues on Windows.

Changes in this CL as below:-
1 Move the plugins sources out of the webkit glue target into a new target 'plugins' which creates a static lib/dll
  depending on the component build flag.
  This target is specified by the newly added webkit_plugins.gypi file in the webkit\plugins folder. This is included
  by webkit_support.gyp. The motivation for moving the plugins out to an independent target is to enable the chrome
  browser code to link to the plugins without depending on glue.

2. We have a plugins_common target in webkit_plugins.gypi which builds a static lib and contains common plugin functionality
   like plugin_utils, webplugininfo, etc. Next step is to move the plugin list sources to this common target as we want to avoid
   browser to depend on Webkit through plugins. That would need some code changes which I wanted to avoid in this CL.

3. The 'glue' target in webkit_glue.gypi currently depends on webkit_media which sadly depends on plugins because of the HTML5
   DRM code. I moved files which are needed by plugins and other consumers like browsers into a static lib target 'glue_common'
   'glue' now depends on 'glue_common' and 'plugins'. 'plugins' depends on 'glue_common'.

4. The other targets like 'renderer', etc also depend on 'glue_common'.

Will move the common plugin files to the newly added webkit common folder in an upcoming CL.

BUG=237249
TBR=cpu@chromium.org, darin@chromium.org, jamesr@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=200375

[darin (slow to review), 2013-05-11 05:26:08]

LGTM

[cpu_(ooo_6.6-7.5), 2013-05-13 19:30:28]

lgtm

[ananta, 2013-05-15 01:04:43]

I looked into the failures on the chrome os builders on the main waterfall when
I landed the earlier incarnation of this change.
http://chromegw.corp.google.com/i/chromium.chromiumos/builders/Linux%20Chromi...

The failures were real and caused by a dependency on plugins introduced via the
webkit_media static lib which is pulled in by glue. The media code depends on
plugins for the HTML5 DRM support. 

To workaround this, I am proposing a change where we split 'glue' into
'glue_common' and 'glue'. 'glue_common' is pulled in by plugins/content/chrome
and 'glue'. 'glue' depends on 'plugins' which should ensure that this change
does not blow up on the waterfall.

PTAL

[jamesr, 2013-05-15 20:42:47]

lgtm

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlreques...
File webkit/glue/weburlrequest_extradata_impl.h (right):

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlreques...
webkit/glue/weburlrequest_extradata_impl.h:18: : NON_EXPORTED_BASE(public
WebKit::WebURLRequest::ExtraData) {
I don't think you should need the NON_EXPORTED_BASE since the derived class
isn't exported

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlrespon...
File webkit/glue/weburlresponse_extradata_impl.h (right):

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlrespon...
webkit/glue/weburlresponse_extradata_impl.h:18: : NON_EXPORTED_BASE(public
WebKit::WebURLResponse::ExtraData) {
ditto

https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
File webkit/plugins/webkit_plugins.gypi (right):

https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
webkit/plugins/webkit_plugins.gypi:51: '<(DEPTH)/base/base.gyp:base_static',
why is this linking in both :base and :base_static? in the component build I
would think base_static should only be linked into base.dll

https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
webkit/plugins/webkit_plugins.gypi:68:
'../../third_party/WebKit/Source/WebKit/chromium/WebKit.gyp:webkit',
why not <(DEPTH)/third_party/... ?

[Jorge Lucangeli Obes, 2013-05-15 20:48:31]

On 2013/05/15 20:42:47, jamesr wrote:
> lgtm
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlreques...
> File webkit/glue/weburlrequest_extradata_impl.h (right):
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlreques...
> webkit/glue/weburlrequest_extradata_impl.h:18: : NON_EXPORTED_BASE(public
> WebKit::WebURLRequest::ExtraData) {
> I don't think you should need the NON_EXPORTED_BASE since the derived class
> isn't exported
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlrespon...
> File webkit/glue/weburlresponse_extradata_impl.h (right):
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlrespon...
> webkit/glue/weburlresponse_extradata_impl.h:18: : NON_EXPORTED_BASE(public
> WebKit::WebURLResponse::ExtraData) {
> ditto
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
> File webkit/plugins/webkit_plugins.gypi (right):
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
> webkit/plugins/webkit_plugins.gypi:51: '<(DEPTH)/base/base.gyp:base_static',
> why is this linking in both :base and :base_static? in the component build I
> would think base_static should only be linked into base.dll
> 
>
https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
> webkit/plugins/webkit_plugins.gypi:68:
> '../../third_party/WebKit/Source/WebKit/chromium/WebKit.gyp:webkit',
> why not <(DEPTH)/third_party/... ?

views_unittests on component linux_chromeos build lgtm

[ananta, 2013-05-15 20:53:34]

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlreques...
File webkit/glue/weburlrequest_extradata_impl.h (right):

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlreques...
webkit/glue/weburlrequest_extradata_impl.h:18: : NON_EXPORTED_BASE(public
WebKit::WebURLRequest::ExtraData) {
On 2013/05/15 20:42:47, jamesr wrote:
> I don't think you should need the NON_EXPORTED_BASE since the derived class
> isn't exported

Done.

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlrespon...
File webkit/glue/weburlresponse_extradata_impl.h (right):

https://codereview.chromium.org/14720012/diff/115030/webkit/glue/weburlrespon...
webkit/glue/weburlresponse_extradata_impl.h:18: : NON_EXPORTED_BASE(public
WebKit::WebURLResponse::ExtraData) {
On 2013/05/15 20:42:47, jamesr wrote:
> ditto

Done.

https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
File webkit/plugins/webkit_plugins.gypi (right):

https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
webkit/plugins/webkit_plugins.gypi:51: '<(DEPTH)/base/base.gyp:base_static',
On 2013/05/15 20:42:47, jamesr wrote:
> why is this linking in both :base and :base_static? in the component build I
> would think base_static should only be linked into base.dll

Done.

https://codereview.chromium.org/14720012/diff/115030/webkit/plugins/webkit_pl...
webkit/plugins/webkit_plugins.gypi:68:
'../../third_party/WebKit/Source/WebKit/chromium/WebKit.gyp:webkit',
On 2013/05/15 20:42:47, jamesr wrote:
> why not <(DEPTH)/third_party/... ?
Cut paste from below:- fixed.

[ananta, 2013-05-15 22:39:51]

Committed patchset #12 manually as r200375 (presubmit successful).