Re: information schema parameter_default implementation

Tom Lane <tgl@sss.pgh.pa.us>

From: Tom Lane <tgl@sss.pgh.pa.us>
To: Peter Eisentraut <peter_e@gmx.net>
Cc: Amit Khandekar <amit.khandekar@enterprisedb.com>, Ali Dar <ali.munir.dar@gmail.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2013-11-17T21:38:45Z
Lists: pgsql-hackers
Peter Eisentraut <peter_e@gmx.net> writes:
> [ 0001-Implement-information_schema.parameters.parameter_de.patch ]

I'm a bit confused as to where this column is coming from?  There's
no such thing in SQL:2008 as far as I can see.  If it's coming from
some not-yet-ratified draft, maybe we should wait for ratification.
It's impossible for a bystander to tell if this implementation conforms
to what the spec is expecting.

BTW, although SQL:2008 lacks this column in the parameters view, there
are about six columns it has that we don't: see the from_sql_xxx and
to_sql_xxx columns.  Surely we should put those in (at least as dummy
columns) before trying to claim adherence to some even-newer spec draft.

As far as the code goes, I have no particular objections, modulo the
question about whether this patch is implementing spec-compatible
behavior.  A small stylistic idea is that maybe the computation of
nth_inputarg should be moved down nearer where it's used.  Really
that's just part of the calculation of nth_default, and it wouldn't
be unreasonable to stick it under the comment explaining why we're
doing that calculation like that.

			regards, tom lane