Doing this resulted in GTK errors about invalid iterators, so postpone the
generation of the CLOSEUP event for slightly later to allow changing the
combobox contents from it with GTK+ 2 (this is not necessary with GTK+ 3).
Also add a demon of a dynamic combobox, creating and destroying its items on
the fly, to the widgets sample.
Closes#17223.
This generic implementation should be used for wxUniv too, otherwise any code
using wxActivityIndicator, such as the widgets sample, simply failed to link
when using it.
Closes#17394.
This warning was given many times and was very confusing as it wasn't given at
all in the right location, but rather for all the other, subsequent
declarations of wxControlRenderer which were correct -- unlike this one which
used wrong order for the "class" keyword and the attribute.
Latest versions of CppUnit removed cppunit-config and use pkg-config instead,
so try the latter if the former is not found to detect CppUnit under Fedora 24
for example.
Closes#17387.
This function replaces gtk_window_set_opacity() and could presumably work
better for the GTK+ versions supporting it.
Also avoid deprecation warnings, if they're ever enabled later, for
gtk_window_set_opacity() which we still have to use for older GTK+.
Closes#17106.
There are just too many of these warnings because GTK+ people are extremely
deprecation-happy and like marking functions which need to be used for the
code to work with the older GTK+ versions as deprecated. Because of this, in
many cases we have no choice but to continue to use the now deprecated
functions and the only way to avoid them is to pepper the code with the
pragmas doing this. Disabling the warning globally seems like the lesser evil
because not disabling them at all is worse than useless as the flood of the
unavoidable deprecation warnings hides any real ones that might occur.
In For the toolbars in non-top orientation, calling SetToolShortHelp() also
showed the item label even if the toolbar was supposed to show the icons only.
Closes#16669.
This makes wxTextCtrl behave like the native text controls and, in particular,
makes it select its entire text when it gets focus (incidentally, this is also
how it behaves under the other platforms).
Closes#9521.
Partially undo the changes of 8e7262fca7: using
wxLogWarning() was wrong, but using asserts is not really great neither as it
results in crashes, and prevents from using, some samples.
Also at least document that this style is not supported.
See #12419.
Positioning wxComboCtrl and wxTextCtrl or wxComboBox in the same column should
align their left borders, but it didn't because of an extra offset for the
focus ring used for wxComboCtrl only.
We need to either apply this offset to all controls or none of them, so remove
it from this one.
Closes#15017.
Add wxGLAttribsBase, wxGLAttributes and wxGLContextAttrs replacing the old
untyped "int attributes[]".
Don't use global object for creating OpenGL > 3.0 contexts.
Closes#16909.
Explicitly bring the parent to top when hiding a modal dialog, this doesn't
seem to do any harm and it fixes a very annoying regression with bringing the
application frame, and not the dialog parent, if it's different, to the top of
Z-order after dismissing the dialog.
Closes#16204.
Due to a bug fixed in 78145f9162 converting a
font which was both underlined and stricken-through to a user string resulted
in using this space-less version and while the bug is fixed now, still accept
the strings created while it was there on input, it doesn't cost much and
results in a better user experience.
See https://github.com/wxWidgets/wxWidgets/pull/187
Honour user selection for these attributes in the native dialog (although only
simple underline/strikethrough are supported currently, not double ones).
See https://github.com/wxWidgets/wxWidgets/pull/187Closes#17338.
Use the scale factor of the associated DC to create the bitmap of the
appropriate size.
This is similar https://github.com/wxWidgets/wxWidgets/pull/158 but uses
portable API instead of wxOSX-specific code.
See #17302.
For a typical scale factor of 2, there won't be any odd-width lines,
and for any factor greater than 1.0, it won't be doing what was intended,
so just don't do it. See #17375
