WITH ... VALUES - Mailing list pgsql-hackers

From Tom Lane
Subject WITH ... VALUES
Date
Msg-id 4685.1266011982@sss.pgh.pa.us
Whole thread Raw
Responses Re: WITH ... VALUES
List pgsql-hackers
Came across something interesting while looking at Marko Tiikkaja's
cut-down WITH patch.  I see that our grammar allows a WITH clause in
front of VALUES, and analyze.c makes some effort to process it, but
AFAICT there isn't any actual use case for this because you can't
reference the WITH clause from the body of VALUES:

regression=# with q as (select * from int8_tbl) values (42);column1 
---------     42
(1 row)

regression=# with q as (select * from int8_tbl) values (q1);
ERROR:  column "q1" does not exist
LINE 1: with q as (select * from int8_tbl) values (q1);                                                  ^
regression=# with q as (select * from int8_tbl) values (q.q1);
ERROR:  missing FROM-clause entry for table "q"
LINE 1: with q as (select * from int8_tbl) values (q.q1);                                                  ^

Even if you go back to 8.4 and turn on add_missing_from, you get:

regression=# with q as (select * from int8_tbl) values (q.q1);
NOTICE:  adding missing FROM-clause entry for table "q"
LINE 1: with q as (select * from int8_tbl) values (q.q1);                                                  ^
ERROR:  VALUES must not contain table references
LINE 1: with q as (select * from int8_tbl) values (q.q1);                                                  ^


So on the whole this seems like useless code.  Perhaps we should instead
throw an error along the line of "WITH cannot be attached to VALUES"?
        regards, tom lane


pgsql-hackers by date:

Previous
From: Arie Bikker
Date:
Subject: Re: xpath improvement V2
Next
From: Arie Bikker
Date:
Subject: Re: xpath improvement V2 with xml.c