Oleksandr Byelkin <sanja@xxxxxxxxxxxxxxxx> writes:
OK, I checked sources of the connector, ad did not see how it can
work, so build and test the test program, it also has the problem (so
user just probably tested old version)
Hehe, I see... I guess connector-c will just need to make the same fix,
then.
Ok, I don't have a better idea than enforcing CLIENT_REMEMBER_OPTIONS when
MYSQL_OPT_NONBLOCK is set. I think the damage was done when
CLIENT_REMEMBER_OPTIONS was disabled originally, but it seems too late to
revert now.
So we just need to document this oddity properly somewhere - along with a
recommendation to always call mysql_close(), even in the case where connect
