On Sun, Aug 14, 2016 at 5:58 AM, Chris Travers <chris.travers@xxxxxxxxx> wrote: > > On Sun, Aug 14, 2016 at 12:13 PM, Xtra Coder <xtracoder@xxxxxxxxx> wrote: >> >> Thanks, I'm aware about ability to create temp functions, but this is >> actually too much overhead - I mean unneeded boilerplate code, but it seems >> in current state it is "the least evil" which I have to use. >> >> I think 'what i need' is support for following >> - ability to switch session language from 'sql' to 'pl/pgsql' >> - in that mode - ability to declare session-scope variables, 'DO' is just >> not needed after that >> - SELECTs not targeted into a variable - are written to client output >> - (C) Merlin Moncure - "Ability to embed collection of statements in the >> database under a name and invoke those statements via CALL <name>, which >> does not automatically create a transaction and a snapshot (unlike >> functions/DO)" >> >> All this seems to be a huge change which will definitely not appear any >> time soon. > > I am willing to bet that DO $$ $$; blocks are neither planned nor > parameterized. And the planner needs to know what is to be returned. The statements within a do block are absolutely planned and parameterized. There was some recent discussion with respect to not planning statements except under certain conditions (in a loop basically) to reduce memory consumption of long 'do' blocks. merlin -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
