Re: [HACKERS] Re: pg_dump possible fix, need testers.

On Mon, Jan 24, 2000 at 12:29:43PM -0500, Tom Lane wrote:
> Patrick Welche <prlw1@newn.cam.ac.uk> writes:
> > Things are still not so good for me. The pg_dumpall > file, psql < file did
> > work, but later:
> 
> > newnham=> select * from crsids,"tblPerson" where
> newnham-> crsids.crsid != "tblPerson"."CRSID";
> > Backend sent B message without prior T
> > D21Enter data to be copied followed by a newline.
> > End with a backslash and a period on a line by itself.
> >>> 
> 
> > which smells like a similar problem. (Note that this is a join. Straight
> > selects didn't cause problems)
> 
> Bizarre.  Obviously, the frontend and backend have gotten out of sync,
> but it's not too clear who's to blame.  Since you also mention
> 
> > (New also, though probably unrelated: the sanity check fails with number of
> > index tuples exactly half number in heap - not equal)
> 
> I think that you may have some subtle platform-specific problems in the
> backend.
> 
> What exactly is your platform/compiler/configuration, anyway?

NetBSD-1.4P/i386 / egcs-2.91.66 19990314 (egcs-1.1.2 release)
configure --enable-debug

This morning I cvs'd made, installed, moved data->data2, initdb, started
up postmaster, reloaded data (this worked!), then tried the join. It's a
big one, so I thought I might as well stress it at the same time, and did
a regression test.

Anything I could try to narrow the problem down?

Cheers,

Patrick