Add more functions for manipulating wxImage colours.
Show using them in the same and add new unit tests for them.
This is a squashed commit of branch 'tomay3000/image-lightness'.
Closes https://github.com/wxWidgets/wxWidgets/pull/2310
Disabling -Wsuggest-override inside macros is broken in gcc, see
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55578, and has started
affecting wxWARNING_SUPPRESS_MISSING_OVERRIDE since gcc 11, i.e. this
macro doesn't have any effect any more and the warning is still given.
Avoid it by actually specifying "override" for gcc 11 (as doing it for
all compilers would result in -Winconsistent-missing-override from
clang) and check that we don't get this warning in the allheaders test.
Also don't use wxDECLARE_ABSTRACT_CLASS() inside wxObject itself, now
that it uses "override", which is not appropriate for the base class
version. This is arguably more clear and should have been done like this
since the beginning anyhow.
Using increment greater than 1 but with a fractional part should still
work, i.e. should use enough digits to fully show the fractional part of
the numbers that can be obtained by using this increment.
wxSpinCtrlDouble::SetIncrement() should increase the number of digits if
necessary because not doing it would be inconsistent with the initial
determination of the number of digits in the ctor and would actually
lose the digits displayed, but it seems unnecessary to decrease the
number of digits and it might be surprising, so don't do it.
Add a test for this behaviour and document it.
This is consistent with using the value of the increment specified in
the ctor for setting the initial number of digits, it was surprising
that creating the control with some value of the increment (e.g. 0.1)
and calling SetIncrement(0.1) later resulted in very different outcomes,
as in the former case the value was shown with a digit after the period
while in the latter case only the integer part was shown.
This also makes the behaviour compatible with that of the previous
versions of the generic wxSpinCtrlDouble, which used "%g" to format the
number before the changes of edc553870f (Fix displaying wxSpinCtrlDouble
values with default precision, 2020-05-18), as they did show the
fractional part even in the latter case.
Add a test checking that this works as expected: before this commit, the
test failed with "1 == 1.2" and "1 == 1.23" errors.
The function was documented to not generate the events, but actually
did generate wxEVT_TEXT ones, even if it didn't generate wxEVT_SPINCTRL.
This was inconsistent with wxGTK and generic wxSpinCtrlDouble used under
MSW, so change this to avoid the unwanted events.
Previously, wxSpinCtrl (using native control) and wxSpinCtrlDouble
(using the generic implementation) behaved differently in this case,
with the former changing its value but the latter keeping the last valid
value instead.
Make them behave the same by resetting the value in both cases and
document this behaviour.
Fix the bug introduced in the recent 34ab87ce4d (Get rid of CppUnit
boilerplate in wxSpinCtrlDouble unit test, 2021-04-19) which resulted in
failures in the subsequent unit tests and ensure that the spin control
is destroyed at the end of the test.
wxSpinCtrl::GetValue() behaviour after calling SetValue("") is currently
inconsistent between wxMSW and wxGTK, or even between wxSpinCtrl and
wxSpinCtrlDouble in wxMSW.
Disable the recently added test checking it for now.
This warning flag is really only used for stdlib debugging/writing,
and is off by default so that normal users of the library don't see
any warnings generated by the library (since they have no control
over it).
Make it possible to use a string value different from the numeric value,
as wxMSW version allows this and some existing code depends on it.
Closes#19140.
This just uses the existing wxImage ctor from XPM data and wxCursor ctor
from wxImage, but will allow the code creating cursors from XPM to still
work even when wxImage ctor from XPM is made explicit.
Add a trivial test just to check that the new ctor can be used.
These paths are actually URLs and so the special URL characters must be
percent-encoded in them.
Document this and add a test checking that this is how it works.
Closes#19142.
Since f7247086c2 ("Fix storing wxBitmap data in GdkPixbuf", 2019-09-18),
919a4ec702 ("Fix drawing wxBitmap with mask", 2019-09-18) and other
commits (see #18498, #18508) RGBA wxBitmaps with masks are drawn properly
under wxGTK2 so if source wxImage has both alpha channel and a mask
the target wxBitmap also should have both components.
Since f7247086c2 ("Fix storing wxBitmap data in GdkPixbuf", 2019-09-18),
919a4ec702 ("Fix drawing wxBitmap with mask", 2019-09-18) and other
commits (see #18498, #18508) RGBA wxBitmaps with masks are drawn properly
under wxGTK2 so only wxBitmap raw RGBA data should be transferred to
wxImage RGBA data because mask is stored in the target wxImage separately.
Make wxAppTraitsBase::GetAssertStackTrace() and reuse it in the assert
handler defined in the test to show more information about the asserts,
especially those failing in worker threads, if possible.
Previously, splitting a string obtained by joining together array
with (any but last) elements ending in the escape character (normally
the backslash), didn't recover the original array because the separator
character following it in the resulting string was considered to be
escaped by wxSplit().
Fix this by escaping the trailing escape character itself.
Add a test confirming that this works as expected now, document this
behaviour and also slightly simplify wxSPlit() logic.
See https://github.com/wxWidgets/wxWidgets/pull/2311Closes#19131.
These tests fail in S/390 Travis CI build, so just don't run them at all
there, as we already did for a couple of tests using /proc that also
failed in other LXC-based builds.
We already did it just before processing the state change event, but
this was too late, as the object could have been already deleted by then
and this actually happened with the example from wxWebRequest
documentation.
Do it earlier now, as soon as the request becomes active, which normally
happens when Start() is called, and keep the reference until the event
is processed after the request reaches one of the final states
(completed, failed or cancelled).
Add a unit test checking that deleting the wxWebRequest object doesn't
prevent the request from running to the completion any more.
Make the code in this function work even with wxFileSystemHandlers that
don't implement FindFirst() at all: we can, and should, still try
loading the passed in path with them, rather than not doing anything at
all.
This makes Load() works with HTTP URLs again, as it apparently used to
do a long time ago, but didn't do any more.
Add a unit test, even if only a semi-manual one, to try to help with
this not getting broken again (ideal would be to launch our own HTTP
server inside the test, but this is a bit more complicated).
Closes#19109.
These tests were also broken by transition to CATCH and didn't run any
longer.
Fix this by replacing CppUnit test suite objects with CATCH test cases
and using a simple helper CheckRE() function instead of creating
complicated test objects hierarchies.
These tests didn't run at all since the changes of e70fc11ef1 (Replace
CppUnit with Catch for unit tests, 2017-11-01) that removed CppUnit:
while they still compiled, the required suite functions were not called
automatically any longer and so nothing happened during run-time.
Finish really removing CppUnit stuff now and just use simple functions
for the checks: this might not be as elegant, but is much simpler and
more transparent.
