also define KDEBINDIR for the tests as it turns out that running
the tests from test target in directory other than the top-level
one (e.g. make -C kdecore/ test) breaks the logic/expectaions in
some tests thus using the full paths to the test tools.
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
these seems unnecessary because QString::fromLocal8Bit() is used
if the function is not overriden, though some sorta UTF-16 support
is suggested in the internal functions I have no trouble creating
files with UTF-16 characters in theirs names and working with
them. if there are any complaints on this it shall be reverted
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
this also fixes a strange bug that I originally tought that it's an RPATH
issues but it turns out that this is some kind of mutex-lock issue. to be
more interesting this bug does not occur when the NDEBUG conditionals are
removed or if they are unconditionalized, the first observation leads me to
think that this is compiler bug and the second that it's mutex-lock issue
that is resolved because some code in KDebug and/or QDebug (i've tried with
it too) locks/unlockes/detaches/does some voodoo that overcomes the bug.
in any case it's hard for me to pin-point the problem so if anyone can look
into it that would be great, for now this change will do
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
by caching results and storing them in a hash the results should
flow faster, a limit should probably be set on this but for now
there isn't. exists() and realPath() now cannot be called without
object!
also, added a debug message to tell when and on what this method
is called. I'm seeing many obscure check done repeatedly,
especially MIME paths check done over and over again when opening
a folder with dolphin so that should help identify places where
optimizations should be made.
for some reason the group is not what the test expects, it may
not fail if the tests are run as root user but I doubt someone
will do that (I do not for one)
