From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andres Freund <andres@anarazel.de>
cc: Alvaro Herrera <alvherre@2ndquadrant.com>,
        Robert Haas <robertmhaas@gmail.com>,
        David Rowley <david.rowley@2ndquadrant.com>,
        PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: pg_dump is broken for partition tablespaces
In-reply-to: <20190422182050.a73zl4wkfwx2ujou@alap3.anarazel.de>
References: 
 <CA+Tgmoa9NsQxwKcqpq-UwYfJ3nYKWFs+p+GYBhq-F_phmUv_VA@mail.gmail.com>
 <20190422181628.GA29329@alvherre.pgsql>
 <20190422182050.a73zl4wkfwx2ujou@alap3.anarazel.de>
Comments: In-reply-to Andres Freund <andres@anarazel.de>
	message dated "Mon, 22 Apr 2019 11:20:50 -0700"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <21096.1555957836.1@sss.pgh.pa.us>
Date: Mon, 22 Apr 2019 14:30:36 -0400
Message-ID: <21097.1555957836@sss.pgh.pa.us>
Precedence: bulk

Andres Freund <andres@anarazel.de> writes:
> On 2019-04-22 14:16:28 -0400, Alvaro Herrera wrote:
>> I think we can get out of this whole class of problems by forbidding the
>> TABLESPACE clause for partitioned rels from mentioning the database
>> tablespace -- that is, users either mention some *other* tablespace, or
>> partitions follow default_tablespace like everybody else.  AFAICS with
>> that restriction this whole problem does not arise, and the patch may
>> become simpler.  I'll give it a spin.

> Why is the obvious answer is to not just remove the whole tablespace
> inheritance behaviour? It's obviously ambiguous and hard to get right.
> I still don't see any usecase that even comes close to making the
> inheritance useful enough to justify the amount of work (code, tests,
> bugfixes) and docs that are required.

Yeah, that's where I'm at as well.  Alvaro's proposal could be made
to work perhaps, but I think it would still end up with some odd
corner-case behaviors.  One example is that "TABLESPACE X" would
be allowed if the database's default tablespace is Y, but if you
try to dump and restore into a database whose default is X, it'd be
rejected (?).  The results after ALTER DATABASE ... SET TABLESPACE X
are unclear too.

			regards, tom lane