Re: Test 002_pg_upgrade fails with olddump on Windows

From: Michael Paquier <michael(at)paquier(dot)xyz>
To: Alexander Lakhin <exclusion(at)gmail(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Test 002_pg_upgrade fails with olddump on Windows
Date: 2023-12-06 01:17:56
Message-ID: ZW_LxBunj-BOMDvT@paquier.xyz
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Tue, Dec 05, 2023 at 12:00:00PM +0300, Alexander Lakhin wrote:
> So removing the condition "if ($oldnode->pg_version != $newnode->pg_version)"
> works here as well, but maybe switching the file mode (to preserve EOLs
> produced by pg_dump) in the block "After dumping, update references ..."
> is more efficient than filtering dumps (on all OSes?).

Well, there's the argument that we replace the library references in
a SQL file that we are handling as a text file, so switching it to use
the binary mode is not right. A second argument is to apply the same
filtering logic across both the old and new dumps, even if we know
that the second dump file taken by pg_dump with not append CRLFs.

At the end, just applying the filtering all the time makes the most
sense to me, so I've applied a patch doing just that.
--
Michael

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Peter Geoghegan 2023-12-06 01:45:41 Re: Bug in nbtree optimization to skip > operator comparisons (or < comparisons in backwards scans)
Previous Message Peter Geoghegan 2023-12-06 01:14:23 Re: Bug in nbtree optimization to skip > operator comparisons (or < comparisons in backwards scans)