Turns out 608fe962cc "Let osl::FileStatus getters
assert programming errors" was overly optimistic and misunderstood that
osl_getFileStatus(..., nMask) /can/ return a FileStatus for which isValid(nMask)
is false, esp. for stuff like file names and URLs of Windows drives and servers.
That in turn leads to existing code now calling
rtl::OUString(_aStatus.ustrFileName) etc. with null argument and crashing.
Change-Id: Icd2168e209aa1c7a6df30cd954513d01034923db
Copied from fpicker/source/win32/filepicker/comptr.hxx, which
otherwise seems to be the same file.
Change-Id: I51ca96c1280e75f9318fcbc74997b2bf755fff95
<http://msdn.microsoft.com/en-us/library/45119yx3%28v=vs.80%29.aspx> "toupper,
_toupper, towupper, _toupper_l, _towupper_l: Visual Studio 2005" states: "In
order for toupper to give the expected results, __isascii and isupper must both
return nonzero." But a community comment clarifies: "This is incorrect, and
should refer to '_toupper' (the macro form), not 'toupper'." (Which makes
sense, as otherwise toupper would violate the C Standard.)
And indeed, at least for some LO built against MSVC 2008 Express on Windows 7,
_toupper('C') = '#' is broken, while toupper('C') = 'C' is good. (And the
unexpected '#' then causes all sorts of errors in the code.)
Change-Id: Iddaddcaf0cc3ffb30e55b0f410a6cfe9118accc8
Remove some dead assignements indicated by clang scan-build report.
Thanks to John Smith clang investigation
Patch corrected
Change-Id: Icbee3a65b6753d45433f1a4ca30567675e27e6d9
...so that backward compatibility ${file::key} -> ${file:key} does not kick in
for the special .link and .override cases.
Change-Id: If009b8ecdbb7662c21e87e7bccb50a0ebbb0190f
i.e. force the __data_start symbol to exist in any executables that link
against libuno_sal so that dlopening of the libgcj provided libjvm.so on some
platforms where it needs that symbol will succeed. e.g. Debian mips/lenny with
gcc 4.3.
With this in place the smoketest succeeds with libgcj provided java. Quite
possibly also required/helpful for s390x/s390 and maybe some others.
Without it the dlopen of libjvm.so will fail with __data_start not found
Change-Id: I01a4de3d7e72d000a8d56a1b9101a187c26a1751
...instead of sal_Bool, to work around MSVC warning C4805: unsafe mix of type
'sal_Bool' and type 'bool'.
Change-Id: Ia3477539ccc23aa9daec49d633f023d414f2b4bf
http://lists.freedesktop.org/archives/libreoffice/2012-April/029940.html
The RTL_USING #define (set by gbuild for anything that's not public
API) allows to use such classes simply by their name, without having
to use the namespace or do explicit using rtl::OUString (which half
of the sources do anyway).
Change-Id: I7edaf12cd278489cdc1d5ff782f0a86361c13c0a
This reverts commit 0b1f424833, which appears to
be a bogus fix (for a bogus warning?). Also needs adaption a la
3c598c4064 "sal: work around spurious signed
overflow warnings."
Conflicts:
sal/inc/rtl/ustring.hxx
Change-Id: I7e9d94fd93a37bb10604c1895ab619a8ac2c3f43
