Horton

Re: WIP: Faster Expression Processing v4

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Cc: pgsql-hackers@postgresql.org, Heikki Linnakangas <hlinnaka@iki.fi>, Doug Doole <ddoole@salesforce.com>
Date: 2017-03-10T19:24:33Z
Lists: pgsql-hackers 
On 2017-03-10 08:51:25 -0500, Peter Eisentraut wrote:
> On 3/7/17 19:14, Andres Freund wrote:
> >> Why shouldn't the function itself also depend on the components of its
> >> return type?
> > Because that'd make it impossible to change the return type components -
> > if the return type is baked into the view that's necessary, but for a
> > "freestanding function" it's not.  If you e.g. have a function that just
> > returns a table's rows, it'd certainly be annoying if that'd prevent you
> > from dropping columns.
> 
> I think functions breaking when the return type is fiddled with is
> actually a not-uncommon problem in practice.  It would be nice if that
> could be addressed.

What problem are you thinking of exactly, and what'd be the solution?
I'd guess that people wouldn't like being unable to change columns in a
table if some function returned the type.

One rather extreme way would be to have functions return a "different"
type, that's initially identical to the table/composite type.  If the
ocmposite type then changes, we'd transparently map between the actual
return type, and the one callers expect.   But that'd a fair bit of
effort, and presumably also quite confusing for users that might
e.g. expect a new column to show up.


> Not necessarily in this patch, but I would like to keep the options
> open.

Yea, seems worth thinking about.  I don't think the patch closes down
options?

Greetings,

Andres Freund

Commits

  1. Improve performance of ExecEvalWholeRowVar.

  2. Remove unreachable code in expression evaluation.

  3. Faster expression evaluation and targetlist projection.

  4. Avoid syntax error on platforms that have neither LOCALE_T nor ICU.

  5. Add configure test to see if the C compiler has gcc-style computed gotos.

  6. Improve regression test coverage for TID scanning.

  7. Improve expression evaluation test coverage.

  8. Fix two errors with nested CASE/WHEN constructs.