> Hmm. It's a feature, but maybe a dangerous one. The expression is > being treated as text(foo), which is intentional in order to allow > use of functions as if they were virtual columns. However, then it > decides that what you've got there is a cast request. There wasn't > any ability to cast composite types to text before 8.3, so this fails > in the expected way in 8.2 and before; but in 8.3 the cast > interpretation succeeds, and away we go. > Thanks for the explanation. I see what's going on now. > foo.char and foo.varchar have similarly unexpected behavior; I think > that's probably the end of it, though, since those are the only types > that CoerceViaIO will take as targets. > ... and also any user defined domains based on those, which is what I actually had. I was unlucky enough that the row text matched the regexp on my domain, so my typo went unnoticed for a while ;-( > Maybe we could/should restrict things so that the syntax continues to > fail, but I can't think of any restrictions that don't seem like warts. > What's worse, they might break stuff that used to work. > > regards, tom lane OK, I can live with that. At least I know what to look out for now! Cheers, Dean _________________________________________________________________ Win New York holidays with Kellogg’s & Live Search http://clk.atdmt.com/UKM/go/111354033/direct/01/
