From: | Rod Taylor <rbt(at)rbt(dot)ca> |
---|---|
To: | PostgreSQL Development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Sequence functions |
Date: | 2003-05-25 03:39:48 |
Message-ID: | 1053833988.279.130.camel@jester |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
We need to track sequence usage in things like defaults. nextval()
function calls don't easily allow this to happen.
The 200N spec has NEXT VALUE FOR <seqname> as the equivalent to
nextval(), which would allow this to happen.
I can make VALUE an IDENT in gram.y to prevent it from becoming a
reserved keyword.
I would also like to add CURRENT VALUE FOR <seqname> for an equivalent
to currval(). CURRENT would need to become a reserved word for this to
happen.
The SQL spec has several reason why CURRENT should be reserved including
several cursor manipulation items (WHERE CURRENT OF), a windowing
function (is this based on cursors?), DISCONNECT, etc.
Any objections to making CURRENT a reserved word?
BTW, VALUE also seems to be used for a special form of unique
constraint. UNIQUE(VALUE) which is equivalent to UNIQUE(SELECT * FROM
<table>).
--
Rod Taylor <rbt(at)rbt(dot)ca>
PGP Key: http://www.rbt.ca/rbtpub.asc
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2003-05-25 03:59:24 | Re: Roadmap for FE/BE protocol redesign |
Previous Message | Bruce Momjian | 2003-05-25 03:19:56 | Re: Roadmap for FE/BE protocol redesign |