"Guillaume Smet" <guillaume.smet@gmail.com> writes:
> I kinda agree with you. The only problem IMHO is that we described in
> the doc exactly what it does and not simply as the big hammer it was
> supposed to be. See
> http://www.postgresql.org/docs/8.3/interactive/sql-discard.html .
Well, the *first* sentence there says it "resets the session to its
initial state", so it seems to me the intent is clear. But maybe we
should alter the second sentence to read, say, "This _currently_ has the
same effect as ...", thereby making it clear that this is implementation
detail and not the controlling definition.
regards, tom lane