Created attachment 274 [details] [review]
gzipped patch

Actually, the README (gtk-nilsb-000506-0.patch.README) says

This replaces gint with gboolean where applicable in exported
functions.

The patch has not yet been applied, and looks mostly correct,
but unfortunately, also includes a lot of occurrences of:

 -  guint          show_text : 1;
-  guint          activity_mode : 1;
+  gboolean      show_text : 1;
+  gboolean      activity_mode : 1;

Which is incorrect, since to hold the values FALSE and TRUE,
a bitfield must be unsigned, and gboolean is signed.

Created attachment 294 [details] [review]
un-gzipped patch

Created attachment 407 [details] [review]
Updated {gint,guint}->gbool patch

Uploaded new and improved patch (removed gboolean ?? : 1, fixed 2
lacunae, and now works against current CVS).

Patch applied, thanks - I think we do want to replace "gint" with
"gboolean" on the event signals in GtkWidgetClass, if you're up for that.

Created attachment 424 [details] [review]
gint->gboolean on event signals in gtkwidget.h

Created attachment 425 [details] [review]
gint->gboolean for event signals in gtkwidget.h and gtkrange.h

Havoc -- uploaded two new patches: one converts gint->gboolean in the
event signals in gtkwidget.h ONLY, and the other does it in BOTH
gtkwidget.h AND gtkrange.h -- I'm not sure which one you wanted.

Range indeed should be fixed. (The need for the change wasn't
limited to gtkwidget.c)

I'm applying your patch and fixing a number of other random
return values for virtual functions, so I think all headers are 
now correct.

There are quite a few event function implementations in the
widgets that return gint and should return gboolean, but
fixing that is really low priority.