Selecting Function Volatility Category - Mailing list pgsql-novice

From Dinesh Somani
Subject Selecting Function Volatility Category
Date
Msg-id CAGcTZwU8J2Te9dnhrWu926FDHOd20LgbxnCz+oqP7=ArJY=P_Q@mail.gmail.com
Whole thread Raw
Responses RE: Selecting Function Volatility Category  (David Raymond <David.Raymond@tomtom.com>)
Re: Selecting Function Volatility Category  (Laurenz Albe <laurenz.albe@cybertec.at>)
List pgsql-novice
I have an id lookup function that translates string names into numeric id. The names are pretty stable over the life of the system but still can mutate once in a while (like, once a month) when someone updates configurations. I can use it like...

  WHERE ... a.attribute_id = f('SALES', 'MATERIAL_NUMBER') ...

Currently I am classfying the function as STABLE. I would prefer to use category IMMUTABLE as that might be more performant. (https://www.postgresql.org/docs/current/xfunc-volatility.html) How does one tackle the occasional case when the cached values of the function no longer apply? Is there some way to trigger cache invalidation?

I had thought of deleting and recreating the function. But that leaves a tiny gap during which the function becomes non-existent (leading to a hit on my SLAs), plus may also cause plan invalidations all over the application.

Regards
Dinesh

pgsql-novice by date:

Previous
From: Stephen Froehlich
Date:
Subject: RE: MySQL to PostgreSQL converter?
Next
From: David Raymond
Date:
Subject: RE: Selecting Function Volatility Category