Postgres Pro
Downloads
Home   >   mailing lists

Thread: RE: [SQL] Column name's length

RE: [SQL] Column name's length

From
"Pham, Thinh"
Date:
I did all of those and still it gave me that problem. What i did was
recompile the whole thing on a diff. machine/install/initdb/createdb...
still it didn't helf. Anyway, i might have to go modify the client program
instead. The only problem is this program is being used by many diff.
companies and if i modify it, and when they upgrade, they're going to have
to modify their database structure too. I just want to minimize the change
as much as possible w/o touching the frontend.

Thanks,
Thinh Pham


-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Tuesday, June 01, 1999 11:27 AM
To: Pham, Thinh
Cc: 'pgsql-sql@postgresql.org'
Subject: Re: [SQL] Column name's length 

You will need to do an initdb --- you are making an incompatible change
in the format of the database.  pg_dump with old code, initdb and reload
with new.

Note that NAMEDATALEN is visible to client applications, which probably
means that libpq, psql, etc, will not interoperate across the two
different settings either.  We're talking rebuild from the ground up here.
        regards, tom lane

Re: [SQL] Column name's length

From
Tom Lane
Date:
"Pham, Thinh" <tpham@mail.priority.net> writes:
>> You will need to do an initdb --- you are making an incompatible change
>> in the format of the database.  pg_dump with old code, initdb and reload
>> with new.

> I did all of those and still it gave me that problem.

Hmm, that probably means that Zalman hasn't found all the dependencies
on NAMEDATALEN after all :-(.  Which version are you using, exactly?

> I just want to minimize the change
> as much as possible w/o touching the frontend.

I think you missed the point I made earlier: this sort of change will
require a frontend recompile because NAMEDATALEN is visible in frontend
apps.
        regards, tom lane

Re: [SQL] Column name's length

From
Zalman Stern
Date:
> Hmm, that probably means that Zalman hasn't found all the dependencies
> on NAMEDATALEN after all :-(.  Which version are you using, exactly?

I just redid my diff and the posted changes really are the only ones to my
6.4.2 sources. This database is working great on a RedHat Linux 5.2 Intel
machine. Without a stack trace from the crashing backend, there's not much
we can do from here...

I did do everything from scratch and I was seeing a crash init'ing the
database before I discovered the PADDED_SBUFDESC_SIZE dependency.

-Z-