Re: 'on insert' rules and defaults - Mailing list pgsql-hackers

From Tom Lane
Subject Re: 'on insert' rules and defaults
Date
Msg-id 28623.955053300@sss.pgh.pa.us
Whole thread Raw
In response to 'on insert' rules and defaults  ("Mark Hollomon" <mhh@nortelnetworks.com>)
List pgsql-hackers
"Mark Hollomon" <mhh@nortelnetworks.com> writes:
> In other words, the default is not honored.

Right, since the INSERT written in the rule provides an explicit
specification of what should be inserted into t.  NEW.b is NULL
and that's what gets inserted.

> I also thought about COALESCE:

> CREATE RULE v_insert AS
>     ON INSERT TO v DO INSTEAD
>     INSERT INTO t values (NEW.i, COALESCE(NEW.b, false));

> But then two places have to know about the default value.

Another problem with that is that there's no way to specify insertion
of a NULL into b.

> Any other suggestions?

You really want default substitution to be done by the parser.
Any later is too late because you won't be able to tell an explicit
NULL from a defaulted column.

I haven't tried it, but I think it would work to declare the "view"
as a real table and then attach the rules to it:

CREATE TABLE t ( i INTEGER,b BOOLEAN DEFAULT false
);

CREATE TABLE v ( i INTEGER,b BOOLEAN DEFAULT false
);

CREATE RULE _RETv ASON SELECT TO v DO INSTEADSELECT * FROM t;

CREATE RULE v_insert ASON INSERT TO v DO INSTEADINSERT INTO t values ( NEW.i, NEW.b);

Then when you do

INSERT INTO v VALUES(43);

the default defined for v.b gets applied by the parser, before the
rule substitution happens.

This still means you have two places that know the default, but
since they're both table declarations maybe it's not so bad.
        regards, tom lane


pgsql-hackers by date:

Previous
From: "Mark Hollomon"
Date:
Subject: 'on insert' rules and defaults
Next
From: Tom Lane
Date:
Subject: Re: postgres crash on CURSORS