Bug 199000 – [GTK] Fix compilation errors for the GTK2 plugin process

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKitGTK

[Adrian Perez]

Build error:

../Source/WebKit/Shared/gtk/WebEventFactory.cpp:104:30: error: assigning to 'guint' (aka 'unsigned int') from incompatible type 'const GdkEventButton' (aka 'const _GdkEventButton')                                                                                             
        eventButton = event->button;
                      ~~~~~~~^~~~~~
../Source/WebKit/Shared/gtk/WebEventFactory.cpp:158:26: error: assigning to 'guint' (aka 'unsigned int') from incompatible type 'const GdkEventButton' (aka 'const _GdkEventButton')                                                                                             
    eventButton = event->button;
                  ~~~~~~~^~~~~~
../Source/WebKit/Shared/gtk/WebEventFactory.cpp:287:21: error: no member named 'keyval' in '_GdkEvent'
    keyval = event->keyval;
             ~~~~~  ^
../Source/WebKit/Shared/gtk/WebEventFactory.cpp:288:22: error: no member named 'hardware_keycode' in '_GdkEvent'
    keycode = event->hardware_keycode;
              ~~~~~  ^
4 errors generated.

[Adrian Perez]

Created attachment 372462 [details]
Patch

[Michael Catanzaro]

Comment on attachment 372462 [details]
Patch

This is a bit weird though, because we are about to delete the plugin process so no more GTK 2, and our minimum GTK version is already 3.6. So it'd be nice to not remove the GTK_API_VERSION_2 guards since that makes it obvious that the code is there to support the plugin process.

I think it's important to investigate what's actually gone wrong here. Clearly something else is up, because #ifndef GTK_API_VERSION_2 and #if GTK_CHECK_VERSION(3, 2, 0) should be equivalent when we know GTK version is either 2 or 3 >= 3.6.

[Adrian Perez]

(In reply to Michael Catanzaro from comment #2)
> Comment on attachment 372462 [details]
> Patch
> 
> This is a bit weird though, because we are about to delete the plugin
> process so no more GTK 2, and our minimum GTK version is already 3.6. So
> it'd be nice to not remove the GTK_API_VERSION_2 guards since that makes it
> obvious that the code is there to support the plugin process.
> 
> I think it's important to investigate what's actually gone wrong here.
> Clearly something else is up, because #ifndef GTK_API_VERSION_2 and #if
> GTK_CHECK_VERSION(3, 2, 0) should be equivalent when we know GTK version is
> either 2 or 3 >= 3.6.

What fixes the issue is the changes in the code, not the difference in
the preprocessor guards themselves. I can re-submit the same fix with
old guards, if you would prefer that :-)

[Michael Catanzaro]

Comment on attachment 372462 [details]
Patch

Oh oops, I even missed that. Yes please.

[Adrian Perez]

Created attachment 372471 [details]
Patch v2

[WebKit Commit Bot]

Comment on attachment 372471 [details]
Patch v2

Clearing flags on attachment: 372471

Committed r246597: <https://trac.webkit.org/changeset/246597>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.