To: "Jeroen T. Vermeulen" <jtv@xs4all.nl>
Cc: pgsql-hackers@postgresql.org, pgsql-sql@postgresql.org
Subject: Re: Cursors and backwards scans and SCROLL 
In-reply-to: <20030310153245.GW98248@xs4all.nl> 
References: <21621.1047242111@sss.pgh.pa.us>
 <20030310153245.GW98248@xs4all.nl>
Comments: In-reply-to "Jeroen T. Vermeulen" <jtv@xs4all.nl>
	message dated "Mon, 10 Mar 2003 16:32:46 +0100"
Date: Mon, 10 Mar 2003 10:41:33 -0500
Message-ID: <17531.1047310893@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

"Jeroen T. Vermeulen" <jtv@xs4all.nl> writes:
>> There are cases where
>> identical cursor definitions might allow or not allow backwards fetch
>> depending on the planner's choices.

> Would it be possible to give warnings in a narrow superset of the
> problematic cases, something along the lines of "I'm scrolling backwards
> for you now, but there's no reason why that should work on this same query 
> tomorrow"?

I don't see a practical way to do that --- that little bit of warning
code would have to embed a lot of fragile assumptions about the set of
alternatives searched by the planner.  It would probably break every
time we improved the planner.  And the breakage would consist either of
failing to give a warning when one is appropriate, or giving a warning
when no other plan is really likely to be chosen; neither of which are
going to be easily noticed or tested for.  Seems like a losing game :-(

			regards, tom lane