Bug 73206 – WebKitGTK+-1.7.2 build error due to a reference to gdk_disable_multidevice()

• Component: WebKitGTK
• Status: RESOLVED FIXED

[Stefan Zwanenburg]

Created attachment 116735 [details]
Build log.

I ran into the following bug on my Gentoo machine: https://bugs.gentoo.org/show_bug.cgi?id=391963. You'll find a full build log as an attachment to that bug (but I'll attach it here anyway).

The problem seems to be that the "testatk" test calls a function that's been introduced in gtk3, which causes an error when WebKit gets built against gtk2. I submitted a patch on the Gentoo bugtracker, which I'll also attach here.

[Stefan Zwanenburg]

Created attachment 116736 [details]
Proposed patch.

[WebKit Review Bot]

Attachment 116736 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files']" exit_code: 1

Total errors found: 0 in 0 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Martin Robinson]

Looks like testatk needs some GTK 2.x love or we should finally decide to drop GTK+ 2.x.

[Mario Sanchez Prada]

(In reply to comment #3)
> Looks like testatk needs some GTK 2.x love or we should finally decide to drop GTK+ 2.x.

I think this problem could be easily solved by just replacing that initializeTestingFramework() function with a regular call to gtk_test_init() + g_setenv("GTK_MODULES", "gail", TRUE), since what that weird function is trying to do is just to ensure the "gail" module is loaded for the unit tests.

Guess now it's when you ask me why in hell I haven't done this first, instead of adding that initializeTestingFramework() function, and the truth is that I just realized about this being a better solution two days ago, while talking to Carlos García Campos... so sorry about that, it was my fault in the beginning.

I will make a patch then and will attach it as soon as I make sure it works fine in the bots. Sorry in advance Stefan for causing you trouble with this and for probably deprecating your patch with the new one I'd be attaching, but if my suggestion works I think it would be a better solution, since it does not rely on copy&paste from GTK, but in a simple call to gtk_test_init()

[Stefan Zwanenburg]

> Sorry in advance Stefan for causing you trouble with this and for probably deprecating your patch with the new one I'd be attaching...

That's no problem at all. Whatever's the best solution should be what ends up being used.

[Mario Sanchez Prada]

(In reply to comment #5)
> > Sorry in advance Stefan for causing you trouble with this and for probably deprecating your patch with the new one I'd be attaching...
> 
> That's no problem at all. Whatever's the best solution should be what ends up being used.

After checking feasibility for my own suggestion I've realized it's not possible since gtk_test_init() internally calls to gtk_init(), so anything we put as GTK_MODULES after calling that function won't work at all.

Hence, your (Stefan) patch it's definitely the way to go.

I will create an proper patch (against trunk and with a ChangeLog entry), preserving your authorship of course, and will upload it for review.

Thanks

[Mario Sanchez Prada]

Created attachment 116788 [details]
Patch

[Martin Robinson]

Comment on attachment 116788 [details]
Patch

Thanks!

[WebKit Review Bot]

Comment on attachment 116788 [details]
Patch

Clearing flags on attachment: 116788

Committed r101271: <http://trac.webkit.org/changeset/101271>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.