> > I just wonder about this statement that you removed: > > * Since we don't want to encourage heavy use of those functions, > > * it seems OK to put a pretty small maximum limit on the number of > > * simultaneously allocated descs. > > > Is it now encouraged to use those functions, or at least that it seems less > > 'scary' than in the past ? > > Well, we could put back some weaker form of the statement, but I wasn't > sure how to word it.
I'm not sure of the 'expected' problems... The only thing I can think of at the moment is the time spent to close file descriptors on abort/commit. I'm not sure of expected value of "max_safe_fds". Your patch now initialize with 5 slots instead of 10, if max_safe_fds is large maybe it is better to double the size each time we need instead of jumping directly to the largest size ? -- Cédric Villemain +33 (0)6 20 30 22 52 http://2ndQuadrant.fr/ PostgreSQL: Support 24x7 - Développement, Expertise et Formation
