David Rowley <dgrowleyml@gmail.com> writes:
> If people don't properly understand these special timestamp input
> values, then maybe the documentation in [1] needs to be improved. At
> the moment the details are within parentheses. Namely "(In particular,
> now and related strings are converted to a specific time value as soon
> as they are read.)". Maybe it would be better to be more explicit
> there and mention that these are special values that the input
> function understands which are translated to actual timestamp values
> when the type's input function is called. That could maybe be tied
> into the DEFAULT clause documentation to mention that the input
> function for constant values is called at DML time rather than DDL
> time. That way, we're not adding these (unsustainable) special cases
> to the documentation.
This sounds like a reasonable approach to me for the
magic-input-values issue. Do we want to do anything about
nextval()? I guess if you hold your head at the correct
angle, that's also a magic-input-value issue, in the sense
that the question is when does regclass input get resolved.
regards, tom lane