Re: Composite type versus Domain constraints. - Mailing list pgsql-general

From Tom Lane
Subject Re: Composite type versus Domain constraints.
Date
Msg-id 29773.1113338933@sss.pgh.pa.us
Whole thread Raw
In response to Composite type versus Domain constraints.  (James Robinson <jlrobins@socialserve.com>)
Responses Re: Composite type versus Domain constraints.  (James Robinson <jlrobins@socialserve.com>)
List pgsql-general
James Robinson <jlrobins@socialserve.com> writes:
> insert into simple_table values (null, '(43)'); -- GRR works!!! It'll
> let any smallint in. What happened to the constraint?

The composite-type input routine doesn't check any constraints ...
and that includes domains.  You can make it work if you don't use
a composite literal:

egression=# insert into simple_table values (null, row(43));
ERROR:  value for domain "simple" violates check constraint "limits"

Obviously this whole area needs improvement.  Domain constraints in
particular fail to be enforced in many places where they should be,
such as plpgsql variables.

            regards, tom lane

pgsql-general by date:

Previous
From: James Robinson
Date:
Subject: Composite type versus Domain constraints.
Next
From: Matt Van Mater
Date:
Subject: psql vs perl prepared inserts