Обсуждение: Re: [PERFORM] Seq scan on zero-parameters function

Dnia 2004-02-06 09:43, Użytkownik Octavio Alvarez napisał:
> Thanks for the hint.
> 
> In fact, my current_period_id() is based on time, but it should be
> constant along the query execution. I mean, I don't want some records
> filtered with some values and other with other values... I'll have an
> uncongruent recordset.

Well - you didn't read the chapter I noticed you, did you?

Look at function now(). It returns always the same value inside 
transaction. If your current_period_id() works the same way as now() 
then declare it as STABLE.

Regards,
Tomasz Myrta

Tomasz Myrta said:
> Dnia 2004-02-06 09:43, U¿ytkownik Octavio Alvarez napisa³:
>> Thanks for the hint.
>>
>> In fact, my current_period_id() is based on time, but it should be
>> constant along the query execution. I mean, I don't want some records
>> filtered with some values and other with other values... I'll have an
>> uncongruent recordset.
>
> Well - you didn't read the chapter I noticed you, did you?

Huummm.. No... :-$

But now I did. Although the chapter makes it look as "how will the
optimizer think the function behaves", not "how the function actually
behaves".

But thanks. It's a lot clearer now. I assume that if I want to make
"timeofday" have a stable-behavior, I must enclose it in a sub-query. Am I
right?

-- 
Octavio Alvarez.
E-mail: alvarezp@alvarezp.ods.org.

Agradezco que sus correos sean enviados siempre a esta dirección.