Thread: Re: Bug#108739: Tablenames should be compiled longer (fwd)

Re: Bug#108739: Tablenames should be compiled longer (fwd)

From
"Oliver Elphick"
Date:
Would anyone like to comment on the advisability or otherwise of
my complying with this request?

------- Forwarded Message

Date:    Wed, 15 Aug 2001 11:25:46 -0500
From:    The Doctor What <docwhat@gerf.org>
To:      Oliver Elphick <olly@lfix.co.uk>
Subject: Re: Bug#108739: Tablenames should be compiled longer


- --a8Wt8u1KmwUX3Y2C
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

* Oliver Elphick (olly@lfix.co.uk) [010814 18:10]:
> The Doctor What wrote:
>   >Package: postgresql-client
>   >Version: 7.1.2-1.1
>   >Severity: normal
>   >Tags: woody
>   >
>   >In src/include/postgres_ext.h about line 39:
>   >
>   >If you change the NAMEDATALEN from 32 to something larger (like 256) it
>   >makes psql work with databases with a larger table size.
>   >
>   >It *should* be okay for only PSQL to be compiled with like this.
>   >
>   >You probably don't want to compile the server with this option due to =
this
>   >warning:
>   >NOTE that databases with different NAMEDATALEN's cannot interoperate!
>
> If the server is not compiled with a longer NAMEDATALEN what is the point
> of doing it with the client? The two are meant to work together.
>
> I suppose you would like to talk to some other server where the length is
> greater, and there is no reason why you should not do that with your
> own system by building from the source.  But please explain why it should
> be done with the official Debian package.

Don't get me wrong, I think it should be expanded for the DB, too.
But the comment implies it'll break people's DBs with a shorter name
length.  Or is it the case that it only breaks if you go *back* to
using a postgres server with shorter lengths.

Your analysis is correct, though, that I'm using a (remote) server
with longer table names.

The reason I thought it worth doing in the client only is that it
can be (and is a lot in the jobs I'm doing) used to talk to remote
databases.   I have found that a lot of the businesses I have delt
with do increase the tablename length to at least 128, usually 256.

Ciao!

- --=20
I'm at two with nature.
     -- Woody Allen

The Doctor What: Kaboom!                         http://docwhat.gerf.org/
docwhat@gerf.org                                                   KF6VNC

- --a8Wt8u1KmwUX3Y2C
Content-Type: application/pgp-signature
Content-Disposition: inline

- -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE7eqKKkJDks3INMZURAoO1AKCfAYIsElvnTT5efkU3GP+FYn4+ywCguixO
dOXGAMnIy1djIQjhFgKfuXc=
=nwsW
- -----END PGP SIGNATURE-----

- --a8Wt8u1KmwUX3Y2C--


------- End of Forwarded Message


--
Oliver Elphick                                Oliver.Elphick@lfix.co.uk
Isle of Wight                              http://www.lfix.co.uk/oliver
PGP: 1024R/32B8FAA1: 97 EA 1D 47 72 3F 28 47  6B 7E 39 CC 56 E4 C1 47
GPG: 1024D/3E1D0C1C: CA12 09E0 E8D5 8870 5839  932A 614D 4C34 3E1D 0C1C
                 ========================================
     "Praying always with all prayer and supplication in the
      Spirit, and watching thereunto with all perseverance
      and supplication for all saints."       Ephesians 6:18



Re: Re: Bug#108739: Tablenames should be compiled longer (fwd)

From
Tom Lane
Date:
"Oliver Elphick" <olly@lfix.co.uk> writes:
> Would anyone like to comment on the advisability or otherwise of
> my complying with this request?

I think it's a lousy idea.  (a) It's not at all clear to me that it's
safe to compile clients with a different NAMEDATALEN from the server.
Even if it happens to be okay with today's sources, the odds of such a
lashup breaking in future are high.  (b) Which NAMEDATALEN are you going
to put in your shipped postgres_ext.h?  Either answer is wrong, since
people might try to use it to compile either frontend or backend code.
(c) I have a very low tolerance for the notion that it's okay for the
Debian distribution to differ however it pleases from what everyone else
ships. That creates support problems for *us*, and so we have a right to
object.

We do have a TODO item to consider raising the standard NAMEDATALEN
value.  So far no one's done any legwork to try to measure space/speed
penalties of larger lengths.

            regards, tom lane

Re: Re: Bug#108739: Tablenames should be compiled longer (fwd)

From
"Oliver Elphick"
Date:
Tom Lane wrote:
  >"Oliver Elphick" <olly@lfix.co.uk> writes:
  >> Would anyone like to comment on the advisability or otherwise of
  >> my complying with this request?
  >
  >I think it's a lousy idea.

I concur; it will not happen.

--
Oliver Elphick                                Oliver.Elphick@lfix.co.uk
Isle of Wight                              http://www.lfix.co.uk/oliver
PGP: 1024R/32B8FAA1: 97 EA 1D 47 72 3F 28 47  6B 7E 39 CC 56 E4 C1 47
GPG: 1024D/3E1D0C1C: CA12 09E0 E8D5 8870 5839  932A 614D 4C34 3E1D 0C1C
                 ========================================
     "Praying always with all prayer and supplication in the
      Spirit, and watching thereunto with all perseverance
      and supplication for all saints."       Ephesians 6:18