PG Bug reporting form <noreply(at)postgresql(dot)org> writes: > postgresql-11.3/src/interfaces/libpq/fe-misc.c:1060]: (style) Comparison of > a function returning boolean value using relational (<, >, <= or >=) > operator. > Source code is > if (forRead && conn->ssl_in_use && pgtls_read_pending(conn) > 0) > but > ./src/interfaces/libpq/libpq-int.h:extern bool pgtls_read_pending(PGconn > *conn);
It looks like the "> 0" belongs inside pgtls_read_pending, since SSL_pending is documented to return an integer count. This is probably harmless with C99-style bools, but if we were doing it like that before PG v12, it could be a live bug in the back branches. I'll check.