Re: Set search_path + server-prepared statements = cached plan must not change result type - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Set search_path + server-prepared statements = cached plan must not change result type
Date
Msg-id 31971.1453301197@sss.pgh.pa.us
Whole thread Raw
In response to Set search_path + server-prepared statements = cached plan must not change result type  (Vladimir Sitnikov <sitnikov.vladimir@gmail.com>)
Responses Re: Set search_path + server-prepared statements = cached plan must not change result type  (Vladimir Sitnikov <sitnikov.vladimir@gmail.com>)
List pgsql-hackers
Vladimir Sitnikov <sitnikov.vladimir@gmail.com> writes:
> Here's my question: why change in search_path does NOT generate
> ParameterStatus message from the backend?

The protocol specification includes a list of the small number of GUCs
for which ParameterStatus messages are generated, and that is not one
of them.

This has been discussed before, I believe, and the conclusion was that
if you think you need this, you're doing it wrong.  More robust solutions
involve having individual functions use SET clauses to locally force
search_path to be what they want.  If we had search_path marked as
GUC_REPORT, that would both lead to a storm of useless client messages
when such techniques were in use, and encourage people to use methods
that won't really work reliably.
        regards, tom lane



pgsql-hackers by date:

Previous
From: Gasper Zejn
Date:
Subject: Proposal for UPDATE: do not insert new tuple on heap if update does not change data
Next
From: Konstantin Knizhnik
Date:
Subject: Re: Batch update of indexes