That's interpretation of subselect is ok, when it contains only stable functions. Maybe add a warning when subselect contains volatile function. 2011/9/2, Rory Campbell-Lange <rory@xxxxxxxxxxxxxxxxxx>: > On 02/09/11, Tom Lane (tgl@xxxxxxxxxxxxx) wrote: >> Rory Campbell-Lange <rory@xxxxxxxxxxxxxxxxxx> writes: >> > I'm doing an UPDATE something like this: >> > UPDATE >> > slots >> > SET >> > a = 'a' >> > ,b = (SELECT uuid_generate_v1()) >> > WHERE >> > c = TRUE; >> >> > Each updated row in slots is getting the same value for b. > >> That's Postgres' interpretation of an uncorrelated sub-SELECT: there's >> no reason to do it more than once, so it doesn't. >> >> > Is there a way of getting a per-row value from uuid_generate_v1() >> > without doing a PL loop? >> >> Drop the word "SELECT". Why did you put that in in the first place? > > Hi Tom > > Good question to which I don't know the answer. Thanks very much for the > advice. > > I was able to force a per-row call to uuid_generate_v1 by using this > pattern > > UPDATE > r_slots > SET b = (SELECT > y.x > FROM > (select -1 as n, uuid_generate_v1() as x )y > WHERE > y.n != r_slots.id) > ... > > But > b = uuid_generate_v1() > is a lot simpler! > > In my "-1" example, am I right in assuming that I created a correlated > subquery rather than an correlated one? I'm confused about the > difference. > > Many thanks > Rory > > > > > -- > Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-general > -- ------------ pasman -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
