Re: PG 7.3 is five years old today

From: Magnus Hagander <magnus(at)hagander(dot)net>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Gregory Stark <stark(at)enterprisedb(dot)com>, Alexandru Cârstoiu <alexandru(dot)carstoiu(at)excelsarc(dot)ro>, pgsql-hackers(at)postgresql(dot)org, "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>
Subject: Re: PG 7.3 is five years old today
Date: 2007-11-28 17:28:54
Message-ID: 474DA556.3080002@hagander.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Tom Lane wrote:
> Gregory Stark <stark(at)enterprisedb(dot)com> writes:
>> Whether there's any need to support the old protocol in the server depends on
>> whether there are any clients out there which use it which is harder to
>> determine and not affected by whether Postgres 7.3 is still around.
>
> Right. There's really not much to be gained by dropping it on the
> server side anyway. libpq might possibly be simplified by a useful
> amount, but on the other hand we probably want to keep its current
> structure for the inevitable v4 protocol.

If we officially remove support for it, we could make modifications to
it without having to consider V2 support. Not that I have any in the
pipeline, but certainly it would make future changes easier if you don't
have to consider backwards compatibility.

Perhaps we could add a warnings message to the logs when a user connects
using the v2 protocol for now, to give users fair warning? (and then
drop it per 8.4). Or to take it even further, a guc that disables
protocol v2 by default but can be enabled for users who are actually
using it?

> Another area where we might think about dropping some stuff is pg_dump.
> If we got rid of the requirement to support dumps from pre-7.3 servers
> then it could assume server-side dependencies exist, and lose all the
> code for trying to behave sanely without 'em.

That would certainly simplify it. There'd still be a supported upgrade
path - just start by upgrading to 8.2 (or really, any supported
version), *then* upgrade from that version to the latest one. That kind
of required-step upgrade is fairly common with commercial products, and
given how old 7.3 is I think it would be very acceptable.

//Magnus

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2007-11-28 17:32:18 Re: pgindent and multiline string constants
Previous Message Joshua D. Drake 2007-11-28 16:57:31 Re: [HACKERS] Time to update list of contributors