wxgtk_initialize_web_extensions() tries to find extensions in the
location where "make install" puts them. If that fails, either because
wx wasn't installed yet, or it was deployed (e.g. in static build)
without them, it looks into additional places - paths relative to CWD.
This is to faciliate running wx samples and tests, but by being enabled
in release builds too, it allowed loading unexpected executable code
from untrusted locations. For example, on typical desktops, one could
put malicious .so files into $HOME/lib to have them loaded by
wxWebView-using applications.
Address this by making the helper paths relative to the executable's
location, not working directory.
Don't log failure to load the extension during initialization, because
it isn't needed for many uses of wxWebView. Instead, only report the
failure at the time when functionality depending on it is used.
Also use g_warning() for logging consistently with other failures in
this file. This doesn't interrupt the user, yet shows the problem highly
visibly in the console.
Change wxGetInstallPrefix() to return a string instead of const
wxChar*. The latter was incorrectly obtained from a temporary string if
WXPREFIX was set. While it's possible to fix in a backward compatible
manner without changing the function's signature, it's not worth the
effort for something pretty obscure and used mostly internally.
wxWidgets already has a mechanism for locating plugins under
$libdir/wx/*, there's no reason to duplicate it with another
compile-time symbol.
As a consequence, this makes the library relocatable via WXPREFIX again.
Fix the location where the web-extension plugin is installed to be
consistent with the primary bakefile-based build system: i.e. to use
lib/wx/x.y.z for devel versions and lib/wx/x.y for stable.
This reverts commit 374db28747 (Fix wxMSW ListCtrl drawing of horizontal
rules for new items, 2016-05-04) and also commit bb3177dd3b (Avoid
infinite repaint loop in wxListCtrl with wxLC_HRULES, 2020-07-24) which
fixed one problem with the original change, but not all of them, as it
was still easily possible to get into an infinite repaint loop.
So just return to the simplest and not too bad, even if not optimal,
solution of refreshing everything after changing a column width when
using horizontal rules.
See #17158, #18850.
Closes#18927.
Make the names more clear and similar to the names of member variables
to which they correspond and also use them consistently in the
documentation.
This reverts the changes to this interface header done in 5925893eed
(Fix various doxygen issues in the documentation, 2019-10-20).
Commit 5925893eed ("Fix various doxygen issues in the documentation")
changed the name to client_data.
This does not fit the naming scheme, and breaks some code in Phoenix
which assumes the param is called clientData (to be able to use a
binding specific type for it).
Rename data to clientData in the inlined overloads. This change is
transparent for any users.
Rename shortHelpString to shortHelp (dito for longHelp). Although this
change is in general visible for any users, Phoenix already mangles the
the parameter names to strip the 'String' suffix.
The introduction of scrollbar animations broke scroll-to-end functionality when
large amounts of text are written, due to interactions with GtkTextView's
background layout. Work around this by scrolling after the layout has finished.
See #18864
The type obviously should be wxWindowID, not int.
Rename the parameter, as it is the only instance where the id is not
named 'id', the implementation already uses 'id', and for Phoenix naming
it id is also preferred, as the name is used for mangling.
The naming change causes no further breakage for Phoenix, as the generator
is currently broken anyway for this case.
Closes https://github.com/wxWidgets/wxWidgets/pull/2069
Fix a bug with wrong size passed to memset() introduced in 1f0ade29f0
(Fix using mask colour even if there is no mask in wxImage::Paste,
2020-09-30) which caused memory corruption and add a test (the one with
the large negative vertical offset) allowing to reproduce this reliably.
Closes https://github.com/wxWidgets/wxWidgets/pull/2067
Verify that constructing wxCaret using its default ctor and Create()
works too, now that it does -- previously it uses to result in a GTK
warning and a crash in wxGTK.
Also update the test to avoid assuming that wxBitmapComboBox inherits
from wxComboBox, which now allows it to build (and pass) on all
platforms.
Closes https://github.com/wxWidgets/wxWidgets/pull/2057
In case an image without alpha is pasted on top of an image with alpha,
the alpha blending gives wrong results. This is caused by the fact that
the final (if nothing has been pasted yet) pixel copying algorithm in
Paste() does not take into account whether the pasted image actually
uses a mask.
To fix this:
- Add the check for image.HasMask().
- In case there is no mask, simply copy the image via memcpy.
- Finally, update the alpha channel of the changed image (if present):
whenever a pixel is copied, the alpha is set to fully opaque.
Closes https://github.com/wxWidgets/wxWidgets/pull/2065
Add the source files present in the bkl but not CMakeLists.txt to the
latter one too.
Also link with AUI library now that a test file using it is included.
Closes https://github.com/wxWidgets/wxWidgets/pull/2064
Current DeviceToLogical{X|Y}(), LogicalToDevice{X|Y}(),
DeviceToLogicalRel{X|Y}(), LogicalToDeviceRel{X|Y}() functions
don't take into account transformations applied with
SetTransformMatrix() so conversion results are invalid if coordinate
system is e.g. rotated.
We need to implement new conversion functions that take into account all
applied transformations and also convert x,y coordinates in one call
because in general case x,y coordinates are coupled and cannot be
converted independently on each other.
Closes#18923.
Generic wxDC::DeviceToLogicalRel{X|Y}() and wxDC::LogicalToDeviceRel{X|Y}()
functions don't take into account scaling applied with
wxDC::SetTransformMatrix().
We need to implement in wxDCImpl and its platform-specific derivates
new conversion functions that take all applied transformations into account.
See #18923.
Fix variable name for the model.
Make the code using wxObjectDataPtr have the same flow as the
code using a raw pointer.
Format the code to be in accordance with the official guidelines.
Closes#18924.
Now that this class is not a template any longer, we can have the code
for radio button group navigation directly in it, without making it
inline, so move the existing functions bodies into the new methods and
remove the old functions entirely.
No real changes, this is just a refactoring.
Use a dirty hack to accommodate wxUniv by deriving wxRadioButtonBase
from wxCheckBox, rather than wxControl, there. This is not pretty, but
should be addressed by refactoring wxUniv code and in the meanwhile all
the other ports don't have to bother with using a template class
unnecessarily.
Now that we do have wxRadioButtonBase class, declare wxRadioButton API
methods as pure virtual in it, to force the derived classes to implement
them.
Also remove the outdated comment saying that there is no base class for
wxRadioButtons in different ports, this is not true any longer.
Allow creating wxWindowPtr from wxScopedPtr<> too, to avoid having to
use .get() in the test code, and add CHECK_SAME_WINDOW() macro which
gives more useful information about the windows in case of failure.
This requires adding a couple of const_cast<>s in their implementation
in order to still allow them returning non-const wxRadioButton pointers,
but this seems preferable to not being able to call them on a const
wxRadioButton in the first place.
