From: Tom Lane <tgl@sss.pgh.pa.us>
To: borisnov@acm.org
cc: pgsql-docs@lists.postgresql.org
Subject: Re: pg_dump no-owner option
In-reply-to: <161052021249.12228.9598689907884726185@wrigleys.postgresql.org>
References: <161052021249.12228.9598689907884726185@wrigleys.postgresql.org>
Comments: In-reply-to PG Doc comments form <noreply@postgresql.org>
	message dated "Wed, 13 Jan 2021 06:43:32 +0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1288536.1610560460.1@sss.pgh.pa.us>
Date: Wed, 13 Jan 2021 12:54:20 -0500
Message-ID: <1288537.1610560460@sss.pgh.pa.us>
Precedence: bulk

PG Doc comments form <noreply@postgresql.org> writes:
> Currently the documentation reads:
> This option is only meaningful for the plain-text format. 

> This statement is imprecise.

Yeah, I agree --- it's not clear whether the option is ignored or
rejected.  Explicitly saying it's ignored is an improvement.

> Would be much better to change the behavior of the utility to allow the
> option for all formats, but htis is somewhat more than just documentaion
> change.

I disagree that that would be an improvement.  Per the next sentence that
you omitted, the right time to apply this and similar options is when
extracting data from the archive format.  (pg_dump to text effectively
constructs an archive dump in-memory and then extracts from that, so that
its "back half" logic is identical to pg_restore's.)  Filtering this info
before dumping the archive would save no meaningful amount of space, while
restricting how the archive could be used later.

			regards, tom lane