Bug 146590

Summary:

[GTK] Disable gtk-doc and gobject-introspection when crossbuilding

Product:

WebKit

Reporter:

Emanuele Aina <emanuele.aina>

Component:

WebKitGTK

Assignee:

Nobody <webkit-unassigned>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

cgarcia, darin, mrobinson, pnormand, svillar

Priority:

Version:

528+ (Nightly build)

Hardware:

OS:

Linux

Attachments:

Description	Flags
Patch	none
Patch	none

Description Emanuele Aina 2015-07-03 06:27:44 PDT

Neither gtk-doc nor gobject-introspection really work when cross-building, let's just disable them when cmake says we're cross-compiling.

Note that in the gtk-doc case, at the moment it is not possible to avoid it even by setting ENABLE_GTKDOC=OFF, since even then it is run as a syntax-check: https://trac.webkit.org/browser/trunk/Source/PlatformGTK.cmake#L35

Comment 1 Emanuele Aina 2015-07-03 06:46:28 PDT

Created attachment 256099 [details]
Patch

Comment 2 Martin Robinson 2015-07-03 07:34:42 PDT

Comment on attachment 256099 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=256099&action=review

> Source/cmake/OptionsGTK.cmake:73
> +# Override the cached variables, gtk-doc and gobject-introspection do not really work when cross-building
> +if (CMAKE_CROSSCOMPILING)
> +    set(ENABLE_GTKDOC OFF)
> +    set(ENABLE_INTROSPECTION OFF)
> +endif ()

Lets move this down to after the USE_LIBHYPHEN check near line 429. It's a bit strange to do this in the middle of the definition of all the options.

Comment 3 Emanuele Aina 2015-07-03 08:01:47 PDT

Created attachment 256105 [details]
Patch

Comment 4 Emanuele Aina 2015-07-03 08:04:40 PDT

> Lets move this down to after the USE_LIBHYPHEN check near line 429. It's a bit strange to do this in the middle of the definition of all the options.

Indeed, I wasn't sure if it was better to keep the override close to the original definition or move it in a more appropriate place.

I've updated the patch and took the liberty of listing you as reviewer in the ChangeLog (is it correct to do so?).

Comment 5 Martin Robinson 2015-07-03 08:12:53 PDT

(In reply to comment #4)
> > Lets move this down to after the USE_LIBHYPHEN check near line 429. It's a bit strange to do this in the middle of the definition of all the options.
> 
> Indeed, I wasn't sure if it was better to keep the override close to the
> original definition or move it in a more appropriate place.
> 
> I've updated the patch and took the liberty of listing you as reviewer in
> the ChangeLog (is it correct to do so?).

It usually isn't necessary. The tools typically fill it in by reading who attaches the r+ in Bugzilla.

Comment 6 Martin Robinson 2015-07-03 08:15:07 PDT

Committed r186259: <http://trac.webkit.org/changeset/186259>

Comment 7 Martin Robinson 2015-07-03 08:16:10 PDT

Thanks for the patch! I took the liberty of landing it.

Comment 8 Emanuele Aina 2015-07-03 08:21:33 PDT

Thanks! :D