Обсуждение: Should we warn users about SETs which have no effect?

Hackers,

There are some GUCs you can set in a context which will accept them, yet
they have no effect when you do.  For example, I can call SET
statement_timeout inside a function, and it has no effect whatsoever.

I'm wondering if we should be throwing a warning in these cases.  And
how many such cases there are.  Thoughts?

--Josh Berkus

On 10/27/09 9:36 AM, Josh Berkus wrote:
> Hackers,
> 
> There are some GUCs you can set in a context which will accept them, yet
> they have no effect when you do.  For example, I can call SET
> statement_timeout inside a function, and it has no effect whatsoever.
> 
> I'm wondering if we should be throwing a warning in these cases.  And
> how many such cases there are.  Thoughts?

Hmmmm, clarification: this applies to SET LOCAL, specifically.

---Josh