Le Vendredi 09 Décembre 2005 09:21, Dave Page a écrit :
> > -----Original Message-----
> > From: Florian G. Pflug [mailto:fgp@phlo.org]
> > Sent: 09 December 2005 01:15
> > To: Dave Page
> > Cc: Andreas Pflug; pgadmin-hackers
> > Subject: Re: [pgadmin-hackers] Schema bug
> >
> > Hm.. couldn't it check the system-schemas by name instead - so, saying
> > "it's a non-system schema if it isn't called information_schema or
> > pg_catalog", instead of "it's a system-schema if oid <
> > LAST_SYSTEM_OID
> > and name is not public"? Or are there other system schemas I
> > don't know
> > of (quite possible ;-) )?
>
> Probably - but it's more of a future-proofing kinda thing. Pretty weak
> argument though...
>
I had two ways to fix this bug : detect system schemas on their names, or
detect public schema by its OID. I think it's better to only keep the public
schema with its OID, it's much less risky to have a wrong one. That's what
the patch is doing :
AND nspname != 'public'
is replaced by
AND nsp.oid != 2200
I still think this patch should be applied.
--
Guillaume.
<!-- http://abs.traduc.org/
http://lfs.traduc.org/
http://traduc.postgresqlfr.org/ -->