Re: Potential bug in ALTER TABLE?

Jeroen Ruigrok/asmodai wrote:

>Hi,
>
>just want to verify first with you guys before dumping it on the bugs
>list. Most likely I am just being silly here or something.
>
>Take this:
>
>create table blah (name TEXT CHECK (name IN ('blah', 'bleh')));
>test=# \d blah
> Table "public.blah"
> Column | Type | Modifiers
>--------+------+-----------
> name | text |
>Check constraints: "blah_name" ((name = 'blah'::text) OR (name = 'bleh'::text))
>
>As we would expect PostgreSQL to do. The constraint has an
>automatically assigned name.
>
>Now, to continue:
>
>ALTER TABLE blah DROP CONSTRAINT blah_name;
>ALTER TABLE blah ADD CHECK (name IN ('blah', 'bleh'));
>test=# \d blah
> Table "public.blah"
> Column | Type | Modifiers
>--------+------+-----------
> name | text |
>Check constraints: "$1" ((name = 'blah'::text) OR (name = 'bleh'::text))
>
>And this time around PostgreSQL doesn't assign an automatic name.
>Well, it depends on what you call a name, but $1, $2, and so on isn't
>quite descriptive. Is this an oversight or am I missing some subtle
>thing here?
>
>
>
You can name it yourself:

ALTER TABLE blah ADD CONSTRAINT blurfl CHECK (name IN ('blah', 'bleh'));

I do this a lot.

I agree the autogenerated names are less than pretty.

cheers

andrew