Bruce Momjian <pgman@candle.pha.pa.us> writes:
> Also, I didn't use palloc because the same C code is used in the backend
> and libpq.
"#define palloc(x) malloc(x)" has been our traditional solution to that.
What I'm more concerned about here is the blithe assumption that a
64-bit-int datatype is available. I'm going through major pushups right
now to ensure that int8 sequences don't break machines without 64-bit
ints, and I'd like to see at least some minimal attention paid to the
issue in this code.
BTW, a protocol version bump for this is a horrid idea. That will
create lots of compatibility problems for people, whether they use
the new auth mode or not.
regards, tom lane