Fix shared library build.

Fix shared library build.

TBR=jam

BUG=90443
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=97511

[jam, 2011-08-19 21:35:13]

http://codereview.chromium.org/7697011/diff/1/content/content_plugin.gypi
File content/content_plugin.gypi (right):

http://codereview.chromium.org/7697011/diff/1/content/content_plugin.gypi#new...
content/content_plugin.gypi:62: '-lXext',
I'm not familiar with this, so I defer to a Linux person

[Evan Martin, 2011-08-19 21:41:01]

http://codereview.chromium.org/7697011/diff/1/content/content_plugin.gypi
File content/content_plugin.gypi (right):

http://codereview.chromium.org/7697011/diff/1/content/content_plugin.gypi#new...
content/content_plugin.gypi:62: '-lXext',
On 2011/08/19 21:35:13, John Abd-El-Malek wrote:
> I'm not familiar with this, so I defer to a Linux person

For this kind of change it's sometimes helpful for a comment call out why these
extra dependencies are necessary.

I guess in this case it's because webplugin_proxy calls Xlib directly.

$ pwd; git grep XShm
/work/chrome/src/content/plugin
webplugin_proxy.cc:    XShmSegmentInfo shminfo = {0};
webplugin_proxy.cc:    *pixmap_out = XShmCreatePixmap(display, root_window,

[Paweł Hajdan Jr., 2011-08-19 21:47:41]

http://codereview.chromium.org/7697011/diff/1/content/content_plugin.gypi
File content/content_plugin.gypi (right):

http://codereview.chromium.org/7697011/diff/1/content/content_plugin.gypi#new...
content/content_plugin.gypi:62: '-lXext',
On 2011/08/19 21:41:01, Evan Martin wrote:
> On 2011/08/19 21:35:13, John Abd-El-Malek wrote:
> > I'm not familiar with this, so I defer to a Linux person
> 
> For this kind of change it's sometimes helpful for a comment call out why
these
> extra dependencies are necessary.
> 
> I guess in this case it's because webplugin_proxy calls Xlib directly.
> 
> $ pwd; git grep XShm
> /work/chrome/src/content/plugin
> webplugin_proxy.cc:    XShmSegmentInfo shminfo = {0};
> webplugin_proxy.cc:    *pixmap_out = XShmCreatePixmap(display, root_window,
> 

I was thinking about a comment, but there are several points in favor of not
doing it:

1) Existing gyp code doesn't comment on those (that'd be a bad excuse though)
2) It is likely to get out of date.
3) Maybe it should rather be something in system.gyp, instead of scattered
randomly throughout all the other gyp files. Like not "fix" the problem by
sticking a comment here, but just cleaning up the general linking with X11.

What do you think?

[Evan Martin, 2011-08-19 21:50:13]

On 2011/08/19 21:47:41, Paweł Hajdan Jr. wrote:
> 2) It is likely to get out of date.
> 3) Maybe it should rather be something in system.gyp, instead of scattered
> randomly throughout all the other gyp files. Like not "fix" the problem by
> sticking a comment here, but just cleaning up the general linking with X11.
> 
> What do you think?

I prefer putting dependencies as close as possible to where they're used.  Most
of chrome shouldn't depend on Xext.  That's also the reason I like comments for
abnormal stuff like this.  (E.g. there is some net test that has a gyp
dependency into chrome/browser somewhere -- a comment really does help)