Re: [HACKERS] seq scan only when function not in subquery (bug?)

From: dg(at)illustra(dot)com (David Gould)
To: brett(at)work(dot)chicken(dot)org (Brett McCormick)
Cc: vadim(at)krs(dot)ru, pgsql-hackers(at)hub(dot)org
Subject: Re: [HACKERS] seq scan only when function not in subquery (bug?)
Date: 1998-06-16 21:40:48
Message-ID: 9806162140.AA11304@hawk.illustra.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> On Tue, 16 June 1998, at 10:32:39, Vadim Mikheev wrote:
>
> > Another issue - handling of functions with constant args
> > in queries - for query
> >
> > select * from T where A = upper ('bbb')
> >
> > function upper ('bbb') will be executed for each tuple in T!
> > More of that - if there is index on T(A) then this index will
> > not be used for this query!
> > Obviously, upper ('bbb') should be executed (by Executor, not
> > parser/planner) once: new Param type (PARAM_EXEC) implemented
> > for subselects could help here too...
> > ---
> >
> > Actually, this is easy to fix...
>
> I was going to reply to this but never did -- how do you tell if it
> needs to be executed once per query or once per tuple? What if you
> wanted to call a function which returned a different value for each
> tuple, like random()?

To make this work, you need an attribute in the functions table (and
internal info about the function) that tells if the function is "variant"
or not. A variant function can return different results with the same
arguments eg random(), or has side effects. A non variant function returns
the same result for the same arguments and has no side-effects.

If you have a non-variant function, then the easy way to optimize it is
to memoize the arguments and result of the last time you called it. Then
the next time you want to call it, check if the arguments are the same and
if so, merely return the previously saved result instead of calling the
function.

Example:

create function city_from_zipcode(integer) returns varchar not variant;

select name, street, city_from_zipcode(zipcode), zipcode
from (select * from customers order by zipcode);

If customers was sorted by zipcode, this would only call city_from_zipcode()
each time the zipcode changed instead of for each row.

It would also cover the case of "function('constant');

-dg

David Gould dg(at)illustra(dot)com 510.628.3783 or 510.305.9468
Informix Software 300 Lakeside Drive Oakland, CA 94612
- A child of five could understand this! Fetch me a child of five.

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message t-ishii 1998-06-17 02:13:11 using a btree index in order by clause?
Previous Message Bruce Momjian 1998-06-16 21:32:59 Re: [HACKERS] non-functional update notice unneccesarily