While it is not guaranteed in general that destroying a window from an
event handler for an event originating from this window itself works, it
did use to work in the case of wxToolBar and its event handlers. However
this stopped working since faffaaae29 as
it added a test using the now deleted object member fields after the
call to the event handler.
To fix this regression without reintroducing the bug fixed by that
commit (see #16762), add an ugly way to check if the toolbar is still
alive after a call to its event handler and do it explicitly after
OnLeftClick() returns.
Closes#17732.
This commit refactors the overloads of cMB2WC() and cWC2MB() methods
taking raw pointers and buffers to reuse the same code and fixes the
wrong length of the buffer returned by cWC2MB(wchar_t*) overload for
conversions using multiple bytes to represent the NUL terminator
character (it previously was wrong for UTF-16 and UTF-32 conversions due
to wrongly subtracting 1 from the length when creating it instead of
correctly subtracting GetMBNulLen()) and the wrong length of the buffer
returned from cMB2WC(char*) overload where no adjustment for the
trailing NUL was done at all.
Also return simple default-constructed buffers from these methods in
case of failure instead of using wxScopedCharBuffer::CreateNonOwned()
which is less obvious and less efficient (even if the latter probably
doesn't matter here because it's only done in case of an error).
Finally, add tests checking that using WC2MB() or either of cWC2MB()
overloads returns the buffers of the same length and with the same
contents.
Instead of adding a separate build, just reuse the existing monolithic
build to also use UTF-8: this is unlikely to miss any problems with the
monolithic build that would be found without using these options, but
also gives us a chance to find the problems in UTF-8 build, both at
compile-time and in the test suite.
Avoid the optimized mb_str() available in this build returning directly
a "char*" as it doesn't preserve the length of the string if it contains
NULs. Use mb_str(wxMBConv) overload instead which always returns the
buffer of the correct length.
Arguably, this is a problem of wxString API and maybe this optimized
mb_str() overload shouldn't be provided at all. OTOH the main reason for
wxUSE_UTF8_LOCALE_ONLY existence is optimization, so it seems a pity to
deliberately make the code less efficient when using it.
The wxGTK implementation uses wx{File,Dir}Button, which derives from
wxButton, but is not a GtkButton, so many operations which try to modify
it, such as changing it's style, are likely to result in errors.
See #17984
We can only use utf8_length() if the conversion used with this object
uses UTF-8 too, otherwise we still need to do the conversion to find out
how many bytes does the string content take in the given encoding.
See #17985.
The Darwin linking problem mentioned in the comment doesn't exist in any
of the still supported macOS versions, so it doesn't make sense to
continue working around it.
This is necessary to properly upgrade the package dependencies when
installing the new versions of the packages themselves and fixes the
problem with the new Python 2.7.14 failing to run since its package was
updated from 2.7.13.
This doesn't work any longer since Cygwin Python 2.7.14 update, see
https://cygwin.com/ml/cygwin/2017-11/msg00053.html
Avoid using it by preventing include/wx/stc/stc.h file from being
updated, this shouldn't be necessary anyhow as the version in git should
be already up-to-date.
Initialize m_pos correctly when using an existing, and hence possibly
not empty, string (and not the internal one which is always empty
initially). The old code was totally wrong as it divided the string
length by the size of wxChar instead of multiplying by it, but doing
this could have been wrong too with UTF-16 and surrogates, so use the
conversion object to compute the real length of the string
representation in the corresponding encoding.
Add a simple unit test checking that this works as intended.
Closes#17985.
This option now lists the available tags instead of showing the timing
information.
We could use Catch "--durations yes" option, but it results in too much
output, so for now just run the tests with default verbosity.
Don't use Catch asserts in threads other than main, this doesn't work
reliably.
Switch to using simple wxASSERT() which doesn't give as much information
as Catch asserts when the test fails, but at least allows the tests to
pass when the asserts don't fail.
This results in an assertion from MSVC CRT implementation which was
somehow consumed by CppUnit but with the switch to Catch resulted in a
test failure.
Drop the legacy CppUnit testing framework used for the unit tests.
Replacing it with Catch has the advantage of not requiring CppUnit
libraries to be installed on the system in order to be able to run
tests (Catch is header-only and a copy of it is now included in the
main repository itself) and, in the future, of being able to write
the tests in a much more natural way.
For now, however, avoid changing the existing tests code as much as
[reasonably] possible to avoid introducing bugs in them and provide
the CppUnit compatibility macros in the new wx/catch_cppunit.h header
which allow to preserve the 99% of the existing code unchanged. Some
of the required changes are:
- Decompose asserts using "a && b" conditions into multiple asserts
checking "a" and "b" independently. This would have been better
even with CppUnit (to know which part of condition exactly failed)
and is required with Catch.
- Use extra parentheses around such conditions when they can't be
easily decomposed in the arrays test, due to the use of macros.
This is not ideal from the point of view of messages given when
the tests fail but will do for now.
- Rewrite asserts using "a || b" as a combination of condition
checks and assert macros. Again, this is better anyhow, and is
required with Catch. Incidentally, this allowed to fix a bug in
the "exec" unit test which didn't leave enough time for the new
process to be launched before trying to kill it.
- Remove multiple CPPUNIT_TEST_SUITE_NAMED_REGISTRATION() macros,
our emulation of this macro can be used only once.
- Provide string conversions using Catch-specific StringMaker for
a couple of types.
- Replace custom wxImage comparison with a Catch-specific matcher
class.
- Remove most of test running logic from test.cpp, in particular don't
parse command line ourselves any longer but use Catch built-in
command line parser. This is a source of a minor regression:
previously, both "Foo" and "FooTestCase" could be used as the name of
the test to run, but now only the latter is accepted.
Don't remove the files in the test cases classes dtors, this was
inconsistent with how they were created: we either need to create the
file in the ctor and destroy it in the dtor or do both only once
globally.
Implement the second solution using the helper AutoRemoveFile instead of
a simple static bool in GetInFileName() implementations.
Use positive tests really checking for whatever we want to check (that
the stream is not seekable in this particular case) instead of just
checking that the test fails, for whatever reason -- which might not be
at all the reason for which we expect it to fail.
It is better to use wholly non-standard wxWidgets smart pointer class
rather than using standard, but deprecated in C++17, std::auto_ptr<> as
this results in warnings from recent compilers.
Because top windows can (and do) act as parents for certain dialogs, a
window already pending delete shouldn't be explicitly reported as the top
window because all dialogs which would use it as a parent
would be destroyed at nearest idle cycle.
Closes#17982.
These files are currently implicitly included from other headers, but we
shouldn't rely on this and include them explicitly as we use std::map<>
and std::auto_ptr<> in this header.
No real changes.
It was only used as part of WXTEST_WITH_GZIP_CONDITION which was
necessary only to support running the tests on systems using zlib < 1.2
which is not a concern since many years any more, so simplify the code
by using the simple non-conditional CPPUNIT_TEST instead and drop the
helper macros which were required only for this.
This is useful when not using libFuzzer (e.g. because the compiler is
gcc or MSVC and not clang) as it allows to debug the problems found by
libFuzzer with the reproducers generated by it.
Including this header as the first wxWidgets header in a translation
unit resulted in compilation errors due to including the internal
wx/wxcrtbase.h header before including wx/defs.h, which must be included
before it (at least to define DLL import/export macros and sized char
types used in wx/wxcrtbase.h).
Fix this by simply including wx/defs.h from wx/buffer.h.
Check for the record size before subtracting it from the end position:
the former must be smaller than the latter for all valid ZIP files and
not performing this check could result in an integer overflow error from
the undefined behaviour sanitizer for bad input.
Credit to OSS-Fuzz: this solves its issue 3828.
Mention that this method is supposed to be used only as an optimization
and also mention that it hadn't been universally available until 3.1.1.
See https://github.com/wxWidgets/wxWidgets/pull/570
Zip filenames containing non ASCII characters will be marked with
bit 11 in the general purpose flags and will use UTF-8 encoding.
By only setting the flag when non ASCII characters are used the
created archives should be binary identical to previous versions.
The old behavior can be achieved by explicitly using wxConvLocal
with the constructor. This should also ensure that
existing code using a custom wxMBConv should work as before.
