No real changes, this is just a refactoring to allow reusing the same
loop waiting until we can get the real window geometry in other tests.
This commit is best viewed with --color-moved.
Using "move" is correct for positioning, as it takes into account the
frame extent. Unfortunately, there is no corresponding API to set the
frame size. So we need to compute the effective client size and use
"resize". We can't use "setGeometry" for this purpose, since a widget's
"geometry" excludes the frame size. We would need to adjust the position
to go from frame to client coordinates. It's more straightforward to
simply have Qt do it with "move".
Closes https://github.com/wxWidgets/wxWidgets/pull/1349
Handle some mouse events explicitly when a wxQtScrollArea is set as
"mouse captured".
The issue arises in that the QScrollArea has two methods: event() and
viewportEvent(). Normally a "QAbstractScrollAreaFilter" is set up by Qt
which routes any events to the viewportEvent() method. And the normal
event() method throws mouse events away (for reasons I'm not aware of -
but it is what it is). If a wx window with a scroll area (e.g.
wxRichTextCtrl) sets capture, the wxQtScrollArea (QScroll-derived) gets
set as the direct "mouse grabber", and all events then bypass the filter
and are sent directly to the "event" method, which throws them away. The
typical result is that the window setting capture no longer gets mouse
events, it keeps capture since it's looking for a mouse up that never
comes, and the app more or less locks up since all mouse events are
being effectively discarded.
This change catches any event that comes in via the event() method and,
when the mouse is captured by the widget, forwards it on to the
viewportEvent method instead, performing what the filter would do via
the normal event routing.
It doesn't forward on "mouse press" events because the initial event
that causes the capture ends up being fed back again, resulting in a
"captured twice" error. The underlying reason I can see for this "being
fed back again" is that, for some inexplicable reason, the
wxRichTextCtrl "skips" the event even though it has actually processed
it and taken capture. That means this solution isn't 100%, but it does
fix the 99%+ cases where the capture is only gotten to redirect mouse
moves and button ups.
Perhaps an alternative solution might be to stop grabbing the mouse in
wxQtScrollArea, but this would require more changes.
Closes https://github.com/wxWidgets/wxWidgets/pull/1346
Building on windowFlags() meant that we inherited a bunch of default
flags that were not affected by Qt::CustomizeWindowHint because we
explicitly specified them when calling setWindowFlags().
Instead of doing this, start with nothing and just add the styles that
we need.
This notably ensures that the frames created without wxCLOSE_BOX style
actually don't have any close button.
Closes https://github.com/wxWidgets/wxWidgets/pull/1335
Avoid calling wxGrid::SetCurrentCell(0, 0) when the grid has no columns
or rows, as it doesn't have any cells then and doing this logically
fails the precondition assert in GetColPos().
Also refactor all 6 different snippets calling SetCurrentCell() in
Redimension() into a single function to simplify the code and make it
more maintainable.
Add a unit test verifying that this works as intended.
Closes https://github.com/wxWidgets/wxWidgets/pull/1546
This allows calling the base class version from the derived classes and
also, at least as importantly, make sure that the function is actually
documented as private functions are not extracted by Doxygen by default.
Closes#16749.
Closes https://github.com/wxWidgets/wxWidgets/pull/1543
Calling wxGrid::{Insert,Delete}{Rows,Cols}() from wxEVT_GRID_CELL_CHANGE
event handler resulted in infinite recursion because it tried to hide
the grid editor control again, which resulted in another CELL_CHANGE
event being generated and so on.
Break this infinite recursion in the usual way, i.e. by updating the
state of wxGrid before invoking the user-defined event handler.
This required separating SaveEditControlValue() in 2 functions, the main
one retaining IsCellEditControlEnabled() check for compatibility, and
the new DoSaveEditControlValue() that can be called even after disabling
the editor.
Closes#2287.
Closes https://github.com/wxWidgets/wxWidgets/pull/1540
Fix handling Enter in multiline text controls in wxMSW.
And other improvements, notably don't send wxEVT_TEXT_ENTER for controls
without wxTE_PROCESS_ENTER style.
Closes#18491.
See https://github.com/wxWidgets/wxWidgets/pull/1539
Since the changes of 04f7f1fd32 (frozen
rows/columns implementation), RefreshBlock() could call GetColPos() with
an invalid index. This didn't matter most of the time as the function
simply returned the same index as long as the columns were using their
natural order, but resulted in a crash due to an out of bound access to
m_colAt array as soon as they were reordered.
Fix this by avoiding using invalid indices in RefreshBlock() and, more
generally, improving its precondition check and making the assumptions
about the input parameters more clear. Also add a defensive check to
GetColPos() itself.
Finally, add a unit test exercising this code.
Closes https://github.com/wxWidgets/wxWidgets/pull/1536
Restore behaviour until c43e0fa123 and let
Enter presses in multiline text controls perform their default function
in the control instead of closing the dialog.
Make the unit test more discerning to check for this.
wxMSW always sent this event to multiline text controls, even when they
didn't have wxTE_PROCESS_ENTER style, contrary to what was documented.
Avoid sending this event unless wxTE_PROCESS_ENTER is used and add unit
tests checking that multiline text controls don't get it without this
style (but still do get it with it).
For once, using the event table macros is preferable because this
bypasses the (generally helpful, but not here) test done by Bind()
verifying that wxEVT_TEXT_ENTER handler is bound to a window with
wxTE_PROCESS_ENTER style.
Doing it like this will allow to check that controls without this style
really do not receive the corresponding event.
This didn't work any more since the changes done in the branch merged by
fb2c17c193 as using wxTE_PROCESS_ENTER
without actually handling the resulting event doesn't prevent Enter from
activating the default button any longer.
Add UpdateText() and UpdateLabel() updating the text (i.e. possibly
containing markup) or the label (just plain text) shown in the window.
Closes#14743.
Closes https://github.com/Kvaz1r/wxWidgets.git BusyInfoUpdateText14743
Update the font of some buddy controls when the DPI changes. Fix the
position of the statusbar after a DPI change. Add some changes that were
suggested in https://github.com/wxWidgets/wxWidgets/pull/1499 but left
out from it.
Some sizes are cached to improve the speed of the library. These sizes
become incorrect when the DPI changes. And are incorrect when a window
is created on a display with a different DPI. Fix this by checking if
the current DPI is the same as the DPI that was used when calculating
the size, otherwise recalculate the size.
Closes https://github.com/wxWidgets/wxWidgets/pull/1530
Ensure that wxPOPUP_WINDOW on Mac always sets the window level
to NSPopupWindowLevel, regardless of the wxSTAY_ON_TOP and
wxFRAME_TOOL_WINDOW flags.
As the comment for wxPOPUP_WINDOW in defs.h says,
"set this flag to create a special popup window:
it will be always shown on top of other windows"
So, that should *always* be true, even in relation to other
Windows with wxSTAY_ON_TOP, since NSPopUpMenuWindowLevel
is greater (higher) than NSModalPanelWindowLevel.
It should also *always* be true even if the window has both
wxPOPUP_WINDOW and wxFRAME_TOOL_WINDOW styles, since there's
no reason why we can't have a panel with NSUtilityWindowMask
and NSPopUpMenuWindowLevel.
For wxImage having both mask and alpha channel (it is technically possible), mask cannot be converted to alpha values and in this case resulting wxBitmap will also have both mask and alpha channel.
Due to lack of support for horizontal scrolling in the native control,
scrolling it was implementing by offsetting the entire control window.
However this didn't work correctly when the window was not positioned
at the leftmost border of its parent window, as the part of it that was
scrolled off could still be visible in this case, and this is exactly
what happened when the native header was used in wxGrid: scrolling it
overwrote the corner part of wxGrid.
Fix this by embedding the actual native control inside an outer wxWindow,
to ensure that the scrolled off part is clipped by the parent window.
Note that this commit is best viewed with "git show --color-moved" as
most of the code was just moved from the header into the implementation
file and is not really new.
