in case the function reference does something strange, that is the case
for pthread_setname_np(), and the return value from the executable is not 0
even tho the source was compiled and linked successfully just return 0
(success) before even attempting to print reference to the function.
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
SIOCGIFNAME and SIOCGIFHWADDR are not documented as appearing in certain
versions of standards or Linux versions thus I assume they should be
defined
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
SQL test segmentation faults and even backtrace is not printed (with unwind
support enabled), not going to debug that
while at it, adjust to changes in Travis config format. no idea what packages
does, not much info at https://config.travis-ci.com/ref/job/cache
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
it has no reliable meaning since QSettings has a system/user scope feature
and it is not guaranteed that settings are actually stored there
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
libpqtypes-dev was required because of the PostgreSQL CMake module and even
with it CMake could not detect it in Travis environment (which is why
custom module is shipped with Katie), libmariadbd-dev does not depend on
libaio-dev even tho it should but that should be fixed in the official
MariaDB Debian package and CMake respectively, I am not going to workaround
those issues anymore just to bump on more
on Debian GNU/Hurd libpqtypes-dev and libaio-dev are not available
(atleast in the officiall main repository) and since they are not
directly required to build Katie I am removing them for that reason too
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
further I/O optimizations may be made via posix_fadvise() and maybe
by getting rid of the write buffer management in QFile and letting
the OS deal with buffering. more research, tests, etc. will have
to be made ofcourse.
on a side note, https://www.gnu.org/software/libc/manual/html_node/Operating-Modes.html
mentions that O_SYNC is another name for O_FSYNC which is a BSD feature
but is currently supported by Linux too. another book reference I found
was in POSIX.4 Programming for Real World by Bill O. Gallmeister, page
281. there it is written that additions are unconditional, if
_POSIX_VERSION is greater or equal than 199309 which is safe to assume
at this point of time I guess.
Signed-off-by: Ivailo Monev <xakepa10@gmail.com>
