Even though it's better to not specify the preview frame size at all,
the size should still be used if it was explicitly specified, but this
didn't happen any more after the addition of the call to Fit().
Fix this now by only doing the equivalent of Fit() if no size was
explicitly given.
Also add advice about not setting the size explicitly to the
documentation.
This was recently broken by bffcb88266 (Optionally detect not filtered
GTK log messages, 2022-05-12), see #22424.
Fix it by providing a trivial "do nothing" version of LogFilterByMessage
even if we can't implement it -- this seems to be preferable to using
preprocessor checks in the code using it.
Closes#22434.
In the future it might be useful to run the code with WXTRACE=gtklog to
see if any GTK log messages we're filtering don't need to be filtered
any longer.
Refactor the code to separate setting of the log callback from the
filtering decision.
Right now the only existing filter is the one just checking the log
level, which is used by GTKSuppressDiagnostics(), but this will allow
adding other filters in the upcoming commits.
Unlike direct tests of __cplusplus, using this macro also works with
(recent enough, i.e. MSVS 2015.3 or later) MSVC versions, even if
/Zc:__cplusplus is not used.
This simplifies some checks and makes some other ones (notably the check
for C++20 used before wxALLOW_COMBINING_ENUMS macro definition) work
with MSVC versions with C++20 support as intended.
There is no need to check for the weird case of using a C++11 compiler
with C++98 standard library any longer, this is not supposed to happen
and we don't support macOS < 10.7 since a very long time anyhow.
This is easy to do and can be useful when using an atomic initialization
counter.
The only platform where not returning the new value might be more
efficient than returning it is Solaris which is not really relevant any
more and on all the other platforms the functions we were already using
provided this value already (or almost, in case of using gcc builtins,
when we just have to use a different one).
Don't offset the DC in the unwanted direction as this resulted in not
drawing anything at all when the grid was scrolled.
Also fix off by one error in the marker line length.
Closes#22403.
Keys such as Ctrl-C etc could be unexpectedly intercepted by
accelerators defined for the menu items when wxSpinCtrl had focus.
Fix this by always preprocessing such keys in wxSpinCtrl itself, just as
it was already done for wxTextCtrl and even reuse the same function for
doing it (which had to be factored out in order to allow it).
Closes#22404.
Allow reusing the code for creating a mask from wxImage mask from
elsewhere.
No real changes yet, this is a just a refactoring.
This commit is best viewed with git --color-moved
--color-moved-ws=ignore-all-space options.
Reuse the same code for handling mouse events for both rows and columns
instead of duplicating almost (but not quite) the same code for both of
them.
As part of resolving the inconsistencies between the two versions, add
wxEVT_GRID_ROW_AUTO_SIZE corresponding to the existing event with the
same name for the columns.
Closes#22380.
Some cleanup in wxDC code: remove unnecessary casts, use RAII helpers
instead of manual memory and other resources management.
No real changes.
See #22378.
Instead of defining a manifest for each individual CPU architecture, use
a single manifest with an asterisk for the architecture.
Should have been part of 77d8926126 (Simplify Windows application
manifests, 2021-12-17).
Closes#22376.
Make wxDC and wxGraphicsContext DPI aware, i.e. add {From,To}DIP() to
them too and make the values returned from Get{DPI,PPI}() consistent
with wxWindow.
Also improve CMake build: add support for finding Cairo when not using
GTK, fix a warning when creating the inplace-config, and mark some
variables as advanced.
See #22346.
Update similar code in all ports to use wxScopedArray for arrays of
points, dashes etc in various wxDC implementations instead of using
new[] and delete[] manually.
No real changes, just make the code safer and shorter.
No real changes, just make the code updating the bounding box slightly
shorter by providing convenient and slightly higher-level overloads.
For now these functions are only in wxDCImpl, it's not clear if we
really need them in wxDC, so don't add them to the public API.
Fix a regression in 3.1.6 which resulted in not showing any images if
the associated image list was empty when it was associated with the
control, as HasImages() returned false in this case and so wxLC_ICON
style was not set.
The fix is to return true from HasImages() if we have image list even if
it is currently empty, as we can't know when (or if) it will become
non-empty and it's better to assume that it will be used (as nothing
really catastrophic should happen if it is not used, finally) rather
than not taking into account the images added to it later.
Closes#22364.
This is slightly more economical, both at run-time and, maybe more
importantly, at compile-time than using wxSharedPtr that we don't really
need here, simple intrusive ref counting smart pointer is enough.
This is not really necessary and breaks encapsulation of this private
method. It is enough to provide just a way to copy the data from another
object as this is all that the derived classes really need.
Add wxSharedClientDataContainer class storing ref-counted client data
and use it instead of plain wxClientDataContainer in wxGridCellAttr,
wxGridCellEditor and wxGridCellRenderer classes.
This allows to keep the same client data associated with many grid cells
without having to make many copies of it.
Don't hardcode inset sizes any longer.
This improves various issues under macOS 12, see #22134 and #22135 (but
doesn't fully fix the former problem yet).
Closes#22351.
It looks like wxGetTranslation() was not actually called for these strings
earlier, so do call it.
The context fulfills also the same purpose as TRANSLATORS: comments --
removing comments as redundant.
