[BUGS] Set-returning functions only allowed if written in language 'sql'

2009-02-10 Thread Daniel Migowski
Hello dear PostgreSQL developers, I noticed the following strange behaviour with set-returning functions. If sets are allowed seems to depend on the language the function is written in, what makes conpletely no sense to me. See the following functions x() and y(). x() is written in 'sql' and

Re: [BUGS] Set-returning functions only allowed if written in language 'sql'

2009-02-10 Thread hubert depesz lubaczewski
On Tue, Feb 10, 2009 at 01:04:02AM +0100, Daniel Migowski wrote: SELECT y(); -- fails with: when you return set, you should use: select * from y(); select function() is additional feature of sql functions. depesz -- Linkedin: http://www.linkedin.com/in/depesz / blog: http://www.depesz.com/

Re: [BUGS] Set-returning functions only allowed if written in language 'sql'

2009-02-10 Thread Pavel Stehule
Hello this limit will be removed at 8.4 in older version you have to use table notation like select * from srf() regards Pavel Stehule 2009/2/10 Daniel Migowski dmigow...@ikoffice.de: Hello dear PostgreSQL developers, I noticed the following strange behaviour with set-returning functions.

Re: [BUGS] Set-returning functions only allowed if written in language 'sql'

2009-02-10 Thread raf
Pavel Stehule wrote: Hello this limit will be removed at 8.4 in older version you have to use table notation like select * from srf() regards Pavel Stehule hi, that's funny. the 8.3 documentation i read said that in the future you'd probably have to use select * from srf() and

Re: [BUGS] Set-returning functions only allowed if written in language 'sql'

2009-02-10 Thread Tom Lane
raf r...@raf.org writes: does this mean that select srf() is no longer deprecated? Well, it's got various issues --- in particular it's not real clear what should happen if there's more than one SRF in a select list. It's unlikely to go away though, especially not before we have an adequate