There is no need to test for the orientation explicitly, as we can just
call AddSpacer() and pass it the spacer size in the direction of the
sizer (which is what we always need).
No real changes, just make the code simpler and nicer to read.
This function used to return bool, but it didn't just always return
true, but it's not even clear how could it possibly fail in theory, so
this was quite useless.
But Realize() itself used GetSize() after each call to RealizeHelper(),
so just make RealizeHelper() itself return this size, which allows to
shorten the code and get rid of a couple of levels of indentation.
No real changes.
This doesn't really change anything because the only existing caller of
SetOrientation() already called Realize() immediately afterwards anyhow
but it seems more logical to do it from SetOrientation() itself, as it
always needs to be done when the orientation changes.
Use the same argument name as in the actual code in the class
constructor as well as the convenience function, instead of
wrongly copypasted "book".
See also 33de6dc (Fix confusing wxPersistentTLW ctor argument name,
2023-12-22).
Closes#24158.
This test doesn't use the native header but somehow still fails on
AppVeyor and locally a very failure can be reproduced if the mouse is
moved while the test runs, so check for this here too.
No real changes, just replace GetEventHandler()->ProcessEvent() with the
calls to this function, which does exactly the same thing, but is
shorter and more clear.
This was only used under Mac but defined under all platforms.
Don't bother with defining it elsewhere and just use a trivial empty
lambda as wxEVT_ACTIVATE handler under Mac.
Note that it's not really clear if this is still needed neither, but it
doesn't cost much to keep this workaround just in case it is.
This does work, contrary to the comment saying that it doesn't: it must
have been when cf6fec73d7 (wxaui flags reworked a bit; allows any
hinting type to be explicity invoked; turning off hinting completely is
not possible; sample updated accordingly, 2006-10-26) was done, but not
any longer.
No real changes, just simplify the code.
Instead of checking for all the other platforms, just check for non-Mac
as the code in this branch should work (and definitely should compile)
in any port and it's just wxOSX which requires special handling here.
This fixes the build under the other platforms (e.g. wxX11) broken by
the changes of e8a6f3b732 (Use wxSYS_COLOUR_HOTLIGHT for hint background
under Mac too, 2023-12-24).
Using 5ms delay for the fade in timer and 4 point increment for the
value running from 0 to 50 meant that the entire fade in took 60ms which
wasn't really noticeable, i.e. couldn't be distinguished from not using
fade in at all.
Increase the delay to 15ms (which corresponds to ~60 FPS) and make the
animation last longer by increasing the transparency more progressively
to make the fade in actually visible.
In the future we probably ought to make all the values involved
(animation duration, initial and end values and the step) configurable
by adding some SetHintFadeInOptions() to wxAuiManager.
Just test whether we're using "Venetian blinds" hint directly instead of
testing for whether we use wxPseudoTransparentFrame which can be also
used for the default transparent hint if transparency is not supported
natively.
This doesn't change much in practice, but just avoid using dynamic cast
unnecessarily.
This makes the hint much more visible under wxGTK where the previously
used wxSYS_COLOUR_ACTIVECAPTION was typically too light in the default
light mode and almost completely invisible in dark mode.
Under wxMSW appearance in dark mode is also improved, although the hint
is still not very visible there -- but this is still better than being
completely invisible before.
Closes#23987.
This reverts commit b1fcf100ff because it
doesn't seem to be necessary any more, as the bug reported in #15071
that was fixed by that commit can't be reproduce any longer in the
current version and this commit resulted in other bugs, notably #24042
and probably also #24063.
The existing code didn't work correctly when wxAUI_NB_CLOSE_ON_ALL_TABS
style was used as it didn't check for it and so assumed that the close
button was only shown on the current page.
This resulted in wrong result being returned from this function for the
right most tab which could not actually fit into the available space
when rendered (because Render() did draw all close buttons correctly)
and, because of this, not showing the tab entirely when MakeTabVisible()
was called, e.g. when adding a new tab.
Closes#24096.
Because some pages in the sample contain too much content to fit entirely
in the visible area of the page. Under wxQt port, this is true regardless
of the state of the window (whether it is maximized or not) some content
remains hidden and inaccessible.
Closes#24154.
Missing semicolon on latest method addition causes doxygen parsing
issues which in turn causes problems for dependent projects like wxRuby
and wxPython.
Closes#24155.
Calling this function from OnInternalIdle() is not always enough, so
call it before processing each and every event to ensure that CEF can
really do whatever it needs to be doing internally.
This doesn't seem to noticeably slow down the program and solves weird
problems, like embedded text boxes (e.g. search zones on the web sites)
not getting focus on click.
wxEventLoop is not used in this file any more after the changes of
11d04389a0 (Add support for using wxWebViewChromium from non-main
thread, 2023-11-29).
Check for 'libcef_dll' instead of 'cef_paths.gypi', because the minimal distribution doesn't contain the last one.
Mark all CEF variables as advanced, so they don't show up in the default CMake GUI.
