On Friday 14 January 2005 17:12, Dave Cramer wrote:
> I see no point in either of these as the solution is simple... Don't
> ignore errors.
This is a misrepresentation of the other side's argument.
I mentioned this thread earlier in this discussion:
http://archives.postgresql.org/pgsql-advocacy/2004-03/threads.php#00067
Consider, in particular,
http://archives.postgresql.org/pgsql-advocacy/2004-03/msg00070.php
The guy is most emphatically _not_ ignoring errors. Wouldn't you
agree?
The point of the solutions that Oliver proposed is not hard to see. I
can write code that works unchanged with Oracle, Sybase, DB2,
MySQL/InnoDB, Firebird and god knows what else. As soon as I throw
PostgreSQL into the mix, I need to handle a radically different
transaction semantics all of a sudden. Oliver's proposal obviates the
need for special-casing PostgreSQL in my application code, albeit
admittedly at the expense of incurring a measurable performance hit.
Which is fine with me, as long as I'm informed of the tradeoff.
YMMV.