srb(at)cuci(dot)nl (Stephen R. van den Berg) writes:
> Anyway, since ctid's solve my problem, I'm not particularly keen on
> getting the LIMIT support on UPDATE/DELETE anymore.
> I still think that the ctid solution is ugly and non-portable.
Well, they're both ugly and nonportable, IMHO.
To me the deciding argument is that the LIMIT approach is also
dangerous, because it encourages people to use incompletely-specified
queries to modify their data. See other messages in this thread for
examples.
regards, tom lane