Hi,
On Mon, Jan 31, 2022 at 06:46:37PM +0100, Pavel Stehule wrote:
>
> I checked the last patch. I think it is almost trivial. I miss just
> comment, why this alias is necessary
>
> + if (!rte->alias)
> + rte->alias = makeAlias(get_rel_name(rte->relid), NULL);
Thanks for looking at it Pavel!
The alias is necessary because otherwise queries involving views won't produce
valid SQL, as aliases for subquery is mandatory. This was part of the v1
regression tests:
+-- test pg_get_query_def()
+SELECT pg_get_query_def('SELECT * FROM shoe') as def;
+ def
+--------------------------------------------------------
+ SELECT shoename, +
+ sh_avail, +
+ slcolor, +
+ slminlen, +
+ slminlen_cm, +
+ slmaxlen, +
+ slmaxlen_cm, +
+ slunit +
+ FROM ( SELECT sh.shoename, +
+ sh.sh_avail, +
+ sh.slcolor, +
+ sh.slminlen, +
+ (sh.slminlen * un.un_fact) AS slminlen_cm,+
+ sh.slmaxlen, +
+ (sh.slmaxlen * un.un_fact) AS slmaxlen_cm,+
+ sh.slunit +
+ FROM shoe_data sh, +
+ unit un +
+ WHERE (sh.slunit = un.un_name)) shoe; +
the mandatory "shoe" alias is added with that change.
I looked for other similar problems and didn't find anything, but given the
complexity of the SQL standard it's quite possible that I missed some other
corner case.