When the original wxGCC_WARNING_SUPPRESS was added, clang understood all
gcc warnings, so it made sense to also apply it when building with
clang, but recent gcc versions have added warnings not available in
clang any more, so we now need a macro for disabling warning the
warnings for gcc only.
Perhaps we should rename the existing wxGCC_XXX macros to use
wxGCC_OR_CLANG prefix.
Due to what looks like a bug, gcc 9.3.0 gives the following incomplete
error message without it:
include/wx/graphics.h:278:7: error: but
‘wxGraphicsGradientStop::wxGraphicsGradientStop(wxGraphicsGradientStop&&)’
does not throw; perhaps it should be declared ‘noexcept’
[-Werror=noexcept]
(without any other diagnostics).
gcc 9 gives -Wnoexcept for these operators and, apparently, not making
them noexcept prevents some optimizations in the standard library
implementation of unordered_foo<>, so do add it.
Unlike the later versions, g++ 4.8 produces a -Wshadow when the name of
a parameter is the same of the name of a method, so rename the (private)
parameters to avoid this.
As wx headers are included from user applications which may compile with
higher warning level than wx itself, try to check headers compilation
with almost all of gcc warning flags turned on.
This notably should prevent the headers from becoming uncompilable with
-pedantic again in the future.
NSTextView doesn't display focus ring by default, which is why wxOSX
did draw it manually, but this behavior can be overriden since OS X
10.3 with NSView.focusRingType property.
The HITheme-based rendering suffered from a number of non-nativeness
issues:
- didn't respect macOS 10.14+ accent colors
- not animated as the native focus ring
- subtly different shape of the outline
- noticeably different outline shape on macOS 11
Remove NeedsFocusRect() and associated workaround for manually drawing
focus ring inside NSTextView (i.e. multiline text controls). This
private interface was only used for wxTextCtrl and nothing else, so
this shouldn't have any impact elsewhere.
Preferences windows traditionally changed their size on macOS when going
between tabs (pages): the window would resize, with animation, to fit the
current tab’s content and avoid wasted space. This worked well, because the
icons toolbar was left-aligned and so changing window size by growing/shrinking
it to the right or down felt natural.
This changes in macOS 11 where preferences toolbar icons are centered and would
shift around if the window resized horizontally. Indeed, the new standard is to
keep the width and only change window height when switching between pages.
Implement this behavior in wxPreferencesEditor too. This requires us to
instantiate all pages early (as is done on other platforms), so this on Big Sur
or newer only.
Extend wxOSXGetSystemImage() and thus wxArtProvider with support for directly
loading the new symbolic images in addition to existing support for named
bundle resources.
This enables usage such as wxArtProvider::GetBitmap("gearshape").
Code compiled against macOS 11 SDK must explicitly specify that the window
should use preferences-style toolbar, otherwise the tabs would be moved to the
right alongside window title, as is done with the standard toolbar now.
e86154fc is insufficient, because there's still one situation when
PostScript names can't be avoided: when storing them using
wxNativeFontInfo::ToString().
Co-authored-by: Stefan Csomor <csomor@advancedconcepts.ch>
50ba73c accidentally omitted an if (colText) check, causing calls to
setTextColor:nil - which in turn caused markup text (but curiously, not
plain) rendering to stick with a previously set color even on cells
where the default color should be used.
Restoring the original logic to prevent setTextColor:nil fixes it.
PostScript name for the system SF font, starting with a dot, was always
considered private, but it was possible to round-trip it. Starting with
macOS 11, font descriptors can't be created from such names and result
in a Times font fallback.
To complicate matters further, the way PostScript name of the system
font is created changed too: it is no longer universal for entire
family, but specifies the weight too.
This combined together makes it impossible to store or modify system
fonts.
Fix by not relying on PostScript names internally. Only use them when
serializing font description or where needed, and in such cases obtain
it from CTFontDescriptor on demand.
Still preserve m_postScriptName as a user-provided detail, and treat it
as such: if provided, it is sacred and used to create font descriptors;
if not provided, never fill it in from other data.
For wxMSW text controls with wxTE_RICH2 style, calling SetFont() counts
as an undoable operation, resulting in CanUndo() returning true even if
no "real" changes have been made yet.
Fix this by resetting the undo stack after creating the control using
ITextDocument::Undo().
Unfortunately this interface is not available in MinGW-32, so this fix
can't be used with it.
Closes https://github.com/wxWidgets/wxWidgets/pull/2010Closes#17524.
If the same function used any of macros from wxCHECK() family more than
once, wxDummyCheckStruct was redeclared, resulting in -Wshadow warnings
from gcc.
Fix this by using unique names for the dummy macro.
Fix a regression introduced in c924ecb10a (Suppress -Wsuggest-override
warnings in user code for gcc too, 2020-07-27) and rearrange the macro
to make sure a semicolon is necessary after it.
Closes#18901.
This changed the type of the art and client ID values, which broke
compatibility with existing code, notably in wxPython (see
https://github.com/wxWidgets/wxWidgets/pull/1996), and the attempts to
fix this compatibility broke it with all the existing code using
wxART_MAKE_ART_ID() or wxART_MAKE_CLIENT_ID() for their own IDs.
Keep things simple and just define the macros as they were defined
before 4552009805 (Merge branch 'pr1312-no-unsafe-wxstring-conv',
2020-07-20), except for wxART_MAKE_CLIENT_ID_FROM_STR() whose argument
and produced value was already of wxString type, and use wxASCII_STR()
at the place of use of wxART_XXX constants in wx code.
This is obviously not ideal as it will require using wxASCII_STR() in
the application code as well, but seems to be the least evil.
This reverts commit 8c9ba23eae, reversing
changes made to 5192feb38e.
Upcoming commits will try to work around the issues with art IDs related
to wxNO_IMPLICIT_WXSTRING_ENCODING in a different way.
Add wx/xrc/xmlres.h to the list of headers compiled with
wxNO_IMPLICIT_WXSTRING_ENCODING to test that they can be used even when
the implicit conversions from "char*" to wxString are disabled.
Simplify the code of wxGrid::DoGridCellLeftDown() and make it more
correct by using m_cursorMode instead of duplicating (not 100%
correctly) the logic used to set it in DoGridMouseMoveEvent().
If the assumption that m_cursorMode is already set by the time
DoGridCellLeftDown() is called turns out to be wrong, e.g. when wxGrid
is used on touch screens, we would need to refactor these functions to
extract the code determining m_cursorMode from the mouse location into a
separate function that would be called from both.
Note that this commit is best viewed with "git diff -w" to see how little
has really changed.
Also add GlobalPtrLock::GetSize() and use it instead of calling
GetSizeFromBuffer() as it's more direct and doesn't require the use of
::GlobalHandle().
Using ::HeapSize() on a global pointer is wrong, and even though it
somehow still works under "genuine" MSW, it crashes under Wine.
Fix this by using ::GlobalSize() instead, which is the right function to
use with this kind of pointer.
Thanks to Damjan Jovanovic for the analysis of the problem in Wine
bugzilla (see https://bugs.winehq.org/show_bug.cgi?id=38924#c10).
Closes#18887.
Preserve the ampersands in the string, which is consistent with wxGTK
behaviour and less surprising than the default behaviour, which creates
mnemonics in the tooltips that are, of course, perfectly useless, as
they can't be activated from keyboard anyhow.
Turn on TTS_NOPREFIX, which is already used by wxToolTip, to fix this.
Closes#18899.
This used to be broken, see #18898, and now that it is fixed by
5a70051c7e (Avoid assertion failure in wxButton with bitmap and empty
label, see #18898, 2020-08-21) add a unit test so that it stays fixed.
Generalize the changes of 415f080c80 (Split wxGrid RANGE_SELECT event
into SELECTING and SELECTED, 2020-07-27) to the case when the mouse is
dragged over row or column headers: also send SELECTING events while
dragging and a SELECTED event at the end, whether it's due to releasing
the mouse button or losing mouse capture.
Change the logic in ChangeCursorMode() to explicitly exclude the modes
for which the mouse should not be captured, as CaptureMouse() should be
called in most cases (and ideally for all of them in the future) and do
capture it for WXGRID_CURSOR_SELECT_{ROW,COL} too, if only to be
notified about mouse capture loss.
Pass the same eventType ExtendCurrentBlock() is called with to
SelectBlock() called to perform the initial selection to ensure that
wxEVT_GRID_RANGE_SELECTING is sent while dragging instead of the
unexpected wxEVT_GRID_RANGE_SELECTED (in addition to the expected one
sent at the end when the drag is over).
Instead of just passing a boolean flag indicating whether
wxEVT_GRID_RANGE_SELECTED should be sent, pass wxEventType to send, with
wxEVT_NULL being interpreted as "don't send anything".
No real changes yet, but this will allow using the existing functions to
send wxEVT_GRID_RANGE_SELECTING and not only SELECTED in the upcoming
commits.
Use the same art provider for a floating frame detached from an existing
wxAuiManager as was used by the original wxAuiManager itself, to ensure
that the appearance of this frame is consistent with the appearance of
its parent.
Implementing this required adding wxAuiDockArt::Clone() to allow copying
it in the new frame and this patch also adds GetAuiManager() to
wxAuiFloatingFrame, similar to the existing method in wxAuiNotebook, in
order to allow changing the dock art from the application code if
desired.
Closes https://github.com/wxWidgets/wxWidgets/pull/2022Closes#18882.
