Bruce Momjian <bruce@momjian.us> writes:
> Robert Haas wrote:
>> The existing behavior is probably not optimal, but I'm not seeing what
>> benefit we get out of neutering it.
> We get to design it right, or maybe not need it at all in 9.1.
Yeah. The good thing about a boolean is that it covers the two
noncontroversial cases (no-wait and wait forever), and doesn't lock
us into supporting cases that we don't really know how to do well
yet.
regards, tom lane