Re: Revisited: Transactions, insert unique.

To All -
O.K., I give! I'm surprised at the number of people willing to dig into
the standards docs, and try to come up with justifications. Note the fact
that whole paragraphs must be quoted in order to get at the general tone
of the standard supports my argument that the behavior on error is _not_
spelled out in the standard: the exact case we're talking about is almost
conspicious by it's absence, given the number of other cases covered,
and the depth of the coverage. The rest of the standard is written in
such away that I keep thinking that all there Exception Conditions must
be defined somewhere, with appropriate actions, but, to the best of my
knowledge, they're not.

Makes me think there must have been a big commercial DB without good
error recovery with representitives on the committee ;-)

Suffice to say, the _right_ thing to do is make the users happy:
now that postgres is more robust, there's little reason not to allow
capture or 'recoverable' errors, and allow the DB developer to decide
when to rollback.

Ross

--
Ross J. Reedstrom, Ph.D., <reedstrm(at)rice(dot)edu>
NSBRI Research Scientist/Programmer
Computer and Information Technology Institute
Rice University, 6100 S. Main St., Houston, TX 77005