Add chrome_split_dll gyp variable, and duplicate chrome_main_dll target

Add chrome_split_dll gyp variable, and duplicate chrome_main_dll target

Currently contains chrome_browser_dll target. Removes some paths from browser->webcore
via content that are there due to single process mode, and removes dependency of content
on content_renderer.

Current remaining paths from chrome_browser_dll to webcore: http://i.imgur.com/7WWRvpA.jpg

R=cpu@chromium.org
BUG=237249
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=198834

[cpu_(ooo_6.6-7.5), 2013-05-01 21:31:25]

lgtm

Lets wait until folks opinionate on the email thread.

[jam, 2013-05-02 18:47:28]

https://codereview.chromium.org/14682014/diff/1/chrome/chrome_dll.gypi
File chrome/chrome_dll.gypi (right):

https://codereview.chromium.org/14682014/diff/1/chrome/chrome_dll.gypi#newcod...
chrome/chrome_dll.gypi:337: # On Windows, link the dependencies (libraries) that
make
There is a lot of duplication below from above. can't you factor out these lists
so that they're in one place and reused?

[scottmg, 2013-05-07 16:40:31]

On 2013/05/02 18:47:28, jam wrote:
> https://codereview.chromium.org/14682014/diff/1/chrome/chrome_dll.gypi
> File chrome/chrome_dll.gypi (right):
> 
>
https://codereview.chromium.org/14682014/diff/1/chrome/chrome_dll.gypi#newcod...
> chrome/chrome_dll.gypi:337: # On Windows, link the dependencies (libraries)
that
> make
> There is a lot of duplication below from above. can't you factor out these
lists
> so that they're in one place and reused?

Removed some of the most obvious duplication, and removed the browser->renderer
dependencies when chrome_split_dll is set (only needed for single process mode).
There's 83 unresolved externals when trying to link chrome_browser.dll
currently.

Also, added a link to graph of the current paths from chrome_browser -> webcore.
Any suggestions on which of those might be removable?

[scottmg, 2013-05-07 18:00:15]

+darin

Does this seem reasonable as a starting point?

[darin (slow to review), 2013-05-07 20:13:19]

LGTM but it would probably be good to get someone more experienced with GYP to
review your GYP changes.

[scottmg, 2013-05-07 20:44:21]

John, would you mind taking a quick look for gyp changes?

It's not exactly pretty, but just somewhere to start. !CHROME_SPLIT_DLL is
around some browser-only things right now where (I guess) we'll have to split
the .cc and gyp targets into a _browser and _child.

[jam, 2013-05-07 20:56:24]

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp
File chrome/chrome.gyp (right):

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp#newcode21
chrome/chrome.gyp:21: 'browser',
why does child depend on browser?

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp#newcode33
chrome/chrome.gyp:33: 'chromium_browser_dependencies': [
i'm a bit confused: why does browser depend on plugin/utility?

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome_dll.gypi
File chrome/chrome_dll.gypi (right):

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome_dll.gypi#ne...
chrome/chrome_dll.gypi:314: 'target_name': 'chrome_browser_dll',
most of this section is still a straight copy from above. can we try to share
that?

[scottmg, 2013-05-07 22:41:03]

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp
File chrome/chrome.gyp (right):

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp#newcode21
chrome/chrome.gyp:21: 'browser',
On 2013/05/07 20:56:24, jam wrote:
> why does child depend on browser?

Haven't started on getting child to actually build yet. Removed that obvious one
though.

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp#newcode33
chrome/chrome.gyp:33: 'chromium_browser_dependencies': [
On 2013/05/07 20:56:24, jam wrote:
> i'm a bit confused: why does browser depend on plugin/utility?

No good reason. Mostly just to get something checked in that links. :/ Removed
those two and #if out of chrome_main_delegate, content_main_delegate, and
content_main_runner.

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome_dll.gypi
File chrome/chrome_dll.gypi (right):

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome_dll.gypi#ne...
chrome/chrome_dll.gypi:314: 'target_name': 'chrome_browser_dll',
On 2013/05/07 20:56:24, jam wrote:
> most of this section is still a straight copy from above. can we try to share
> that?

Mostly removed since it wasn't really necessary anyway for now. Stupid gyp
grumble grumble.

[jam, 2013-05-07 23:08:16]

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp
File chrome/chrome.gyp (right):

https://codereview.chromium.org/14682014/diff/25001/chrome/chrome.gyp#newcode21
chrome/chrome.gyp:21: 'browser',
On 2013/05/07 22:41:03, scottmg wrote:
> On 2013/05/07 20:56:24, jam wrote:
> > why does child depend on browser?
> 
> Haven't started on getting child to actually build yet. Removed that obvious
one
> though.

it's still there?

https://codereview.chromium.org/14682014/diff/35001/chrome/chrome_dll.gypi
File chrome/chrome_dll.gypi (right):

https://codereview.chromium.org/14682014/diff/35001/chrome/chrome_dll.gypi#ne...
chrome/chrome_dll.gypi:350: 'CHROME_DLL',
actually, looks like these defines aren't used anywhere in the code. can you
take them out from here and above? i think these are long-ago leftovers from a
previous attempt to have a dll per process. looks like they got added in
https://codereview.chromium.org/125021

[scottmg, 2013-05-07 23:14:15]

Really removed browser now.

https://codereview.chromium.org/14682014/diff/35001/chrome/chrome_dll.gypi
File chrome/chrome_dll.gypi (right):

https://codereview.chromium.org/14682014/diff/35001/chrome/chrome_dll.gypi#ne...
chrome/chrome_dll.gypi:350: 'CHROME_DLL',
On 2013/05/07 23:08:16, jam wrote:
> actually, looks like these defines aren't used anywhere in the code. can you
> take them out from here and above? i think these are long-ago leftovers from a
> previous attempt to have a dll per process. looks like they got added in
> https://codereview.chromium.org/125021

Done.

[jam, 2013-05-07 23:21:20]

thanks, lg

one question: you're using the CHROME_SPLIT variable now to really mean
CHROME_SPLIT_THIS_IS_THE_BROWSER_DLL_PART (not real naming, just to illustrate
what i mean). you'll need to change this when you build the child dll, so maybe
start with the real one now?

[scottmg, 2013-05-07 23:26:01]

On 2013/05/07 23:21:20, jam wrote:
> thanks, lg
> 
> one question: you're using the CHROME_SPLIT variable now to really mean
> CHROME_SPLIT_THIS_IS_THE_BROWSER_DLL_PART (not real naming, just to illustrate
> what i mean). you'll need to change this when you build the child dll, so
maybe
> start with the real one now?

yes, that's exactly what it is right now (in some places).

i'm not sure how we want to do that, i don't think we want a define (for that
purpose at all).

i was thinking that we'll need to split those files into two pieces so that one
can go in a gyp lib that's blah_browser and the other that's in blah_child. but
yes, that's what i was planning on doing next.

[commit-bot: I haz the power, 2013-05-08 00:14:55]

Change committed as 198834