New.* and old.* as function arguments within rules

Karl O. Pinc <kop@meme.com> schrieb:

Hi,

I'm trying to make sure I understand what I'm doing.

Where is new.* and old.* documented, as regards
using them as arguments to functions called from
rules? If it's not documented then can I rely

http://www.postgresql.org/docs/8.1/interactive/triggers.html

HTH Andreas
--
Really, I'm not out to destroy Microsoft. That will just be a completely
unintentional side effect. (Linus Torvalds)
Kaufbach, Saxony, Germany, Europe. N 51.05082�, E 13.56889�

On 12/03/2005 01:43:38 AM, Andreas Kretschmer wrote:

Karl O. Pinc <kop@meme.com> schrieb:

Hi,

I'm trying to make sure I understand what I'm doing.

Where is new.* and old.* documented, as regards
using them as arguments to functions called from
rules? If it's not documented then can I rely

on the behavior?

http://www.postgresql.org/docs/8.1/interactive/triggers.html

Thanks for the reply but I've obviously got some big
misunderstanding here.

1st, the docs above refer to triggers, not rules. AFIK
rules are a completely different animal.

2nd, nowhere have I found a NEW.* syntax (as written).
NEW (or OLD) seems to be a complete rowtype (or maybe
recordtype), as far as triggers go anyway, and there's
no explaination of what .* might mean in the context
of a rowtype.

Finally, the syntax seems to have something to do
with calling functions.

See: http://archives.postgresql.org/pgsql-general/2005-08/msg00653.php
Which is what I found searching the archives trying to
figure out the best way to pass data in NEW and OLD
to functions called from within rules.

Karl <kop@meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein

"Karl O. Pinc" <kop@meme.com> writes:

2nd, nowhere have I found a NEW.* syntax (as written).

This could certainly stand to be better documented, but there is an
example for instance here:
http://developer.postgresql.org/docs/postgres/xfunc-sql.html#AEN31568

In general, "foo.*" where foo is a visible table alias is meaningful
anywhere that a rowtype value would be accepted. There is a special
case at the top level of a SELECT result list, where it will be broken
apart into a list of foo's component fields because the SQL spec says
so.

At one time you could just write "foo" instead of "foo.*", but that's
deprecated because it's ambiguous against the case of a simple column
"foo". I'm not sure to what extent it still works at all, and it
probably will stop working in any remaining cases someday.

NEW/OLD are not different from other table aliases as far as these
matters go.

regards, tom lane

On 12/03/2005 10:29:43 PM, Tom Lane wrote:

"Karl O. Pinc" <kop@meme.com> writes:

2nd, nowhere have I found a NEW.* syntax (as written).

In general, "foo.*" where foo is a visible table alias is meaningful
anywhere that a rowtype value would be accepted. There is a special
case at the top level of a SELECT result list, where it will be broken
apart into a list of foo's component fields because the SQL spec says
so.

Thanks very much. I had just started writing 'foo' because I didn't
know about 'foo.*' and it's been working so far (8.0.3) but
would _hate_ to have it break on upgrade.

This is the first time I've found the Postgres documentation to be
really lacking. I suppose there has to be a first.
(Very sorry, _wway_ to busy to write a doc patch.)

Karl <kop@meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein