On 10.11.21 16:41, Tom Lane wrote:
Peter Eisentraut <pe...@eisentraut.org> writes:
Remove check for accept() argument types
Early returns from the buildfarm are
gaur | 2021-11-09 16:55:58 | auth.c:3235:17: warning: pointer
targets in passing argument 6 of 'recvfrom' differ in signedness
gaur | 2021-11-09 16:55:58 | pqcomm.c:722:9: warning: pointer
targets in passing argument 3 of 'accept' differ in signedness
gaur | 2021-11-09 16:55:58 | pqcomm.c:743:6: warning: pointer
targets in passing argument 3 of 'getsockname' differ in signedness
gaur | 2021-11-09 16:55:58 | pgstat.c:483:39: warning: pointer
targets in passing argument 3 of 'getsockname' differ in signedness
gaur | 2021-11-09 16:55:58 | pgstat.c:630:9: warning: pointer
targets in passing argument 5 of 'getsockopt' differ in signedness
gaur | 2021-11-09 16:55:58 | fe-connect.c:2760:11: warning: pointer
targets in passing argument 5 of 'getsockopt' differ in signedness
gaur | 2021-11-09 16:55:58 | fe-connect.c:2788:9: warning: pointer
targets in passing argument 3 of 'getsockname' differ in signedness
Right offhand I don't see any other animals complaining.
May I suggest that "unsigned int" would be a better choice
than "int" for socklen_t?
I have been waiting for a few more buildfarm members to finish (mainly
the other AIX and HPUX instances), but they appear to be on strike right
now. But based on existing results and extrapolation, it might be that
gaur is actually the only one without socklen_t, so we can do whatever
we want to make it happy. What does the man page say the correct type
would be? size_t?
