From: Josh Berkus <josh@agliodbs.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
To: pgsql-hackers@postgreSQL.org
To: pgsql-sql@postgreSQL.org
Subject: Re: [SQL] Cursors and backwards scans and SCROLL
Date: Sun, 9 Mar 2003 12:59:35 -0800
Message-ID: <200303091259.35423.josh@agliodbs.com> (raw)
In-Reply-To: <21621.1047242111@sss.pgh.pa.us>
References: <21621.1047242111@sss.pgh.pa.us>
Tom,
> Postgres' implementation of cursors has always had a problem with doing
> MOVE or FETCH backwards on complex queries.
Coincidnetally enough, I was just chatting with one of my contractors
yesterday about how the one thing that Transact-SQL has to offer is a really
good cursor implementation. It would be lovely to improve ours to match.
> Fixing this directly seems unreasonably difficult, so I'm currently
> working on fixing it by inserting a Materialize plan node at the top of
> the plan tree for a cursor, if the plan tree couldn't otherwise support
> backwards scan. The Materialize node will save aside a copy of each row
> as it's fetched from the underlying plan, and use this copied table if
> any backwards scanning is asked for.
Sounds good to me. It's also very similar to what T-SQL does for a STATIC or
KEYSET cursor, and works very well in their implementation. (FWIW, T-SQL's
cursor types, such as DYNAMIC and KEYSET, are unnecessary for Postgres due to
MVCC)
> 2. Error out only if a backwards fetch is actually attempted on a plan
> tree that can't handle it (which could only happen if SCROLL wasn't
<snip>
> I'm presently leaning to #2, even though it exposes implementation
> details. I'm open to discussion though. Any preferences? Other ideas?
This sounds like a good idea to me in a staggered-implementation sense if it's
doable. That is, we'd implement the behavior in #2 in the next version of
Postgresql, and the behavior in #1 or in #3 in the version after that. If,
however, the implementation of #2 is too difficult, then I think #3 would be
a good choice.
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-sql@postgresql.org
Cc: josh@agliodbs.com, tgl@sss.pgh.pa.us, pgsql-hackers@postgreSQL.org
Subject: Re: [SQL] Cursors and backwards scans and SCROLL
In-Reply-To: <200303091259.35423.josh@agliodbs.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox