Re: Could regexp_matches be immutable?

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Peter Eisentraut <peter_e(at)gmx(dot)net>
Cc: Rod Taylor <rod(dot)taylor(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Could regexp_matches be immutable?
Date: 2009-10-21 13:12:31
Message-ID: 26630.1256130751@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> On Tue, 2009-10-20 at 20:48 -0400, Tom Lane wrote:
>> So I went to see about making the changes to remove regex_flavor, and
>> was astonished to find that all the regex-related functions are already
>> marked immutable, and AFAICS always have been. This is clearly wrong,
>> and we would have to fix it if we weren't about to remove the GUC.

> Are you sure this wasn't intentional, because it breaks performance and
> we doubted that many applications would change regex_flavor on the fly?

Intentional or not, it's wrong :-(

In practice I doubt there are many cases where constant-folding a regex
would be possible or performance-critical. The real use of having it
be immutable is probably Rod's, ie, using it in an index. And that is
*obviously* really dangerous if there's a GUC affecting the results.

regards, tom lane

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Alvaro Herrera 2009-10-21 13:28:18 Re: \du quite ugly in 8.4
Previous Message Andrew Dunstan 2009-10-21 13:08:23 Re: UTF8 with BOM support in psql