Karsten Hilbert <Karsten(dot)Hilbert(at)gmx(dot)net> writes:
> Also a consideration: table.*::text may become quite unwieldy
> if there's one or more BYTEA columns in the table.
One other thought here --- if you do want to go with the "no other
updates" semantics, it still seems like it should be sufficient to
compare xmins. Comparing the xmax values would add nothing to that,
except that it would reject if another update had been attempted and
then failed, which seems undesirable.
regards, tom lane