Re: vacuum_truncate configuration parameter and isset_offset - Mailing list pgsql-hackers

From David G. Johnston
Subject Re: vacuum_truncate configuration parameter and isset_offset
Date
Msg-id CAKFQuwb2_KrsduY6+ZcWri2ftNw6=EggAQVPrO630OaoANzAag@mail.gmail.com
Whole thread Raw
In response to Re: vacuum_truncate configuration parameter and isset_offset  (Nikolay Shaplov <dhyan@nataraj.su>)
List pgsql-hackers
On Wed, Mar 26, 2025 at 10:29 AM Nikolay Shaplov <dhyan@nataraj.su> wrote:
The only thing I am asking for: please keep code consistent. Better keep
reloption core the way it was, but if it is not possible, then keep it
consistent then.

Consistency at the expense of all else is a debatable position that will benefit from being actively justified.  Incremental improvement has a lot of merit.  The cost of incremental improvement is quite low for generally high benefit.  Wholesale rewriting has a much higher cost for basically the same benefit.  That's a lot of added cost in order to stay "consistent", so what are the added benefits of consistency for incurring that cost?

There is a baseline benefit to consistency that any incremental improvement needs to overcome to be worthwhile.  I don't think the isset_offset implementation to handle this rare boolean option overcomes that baseline.  I've tossed my weight, expect others to do so as well, and we'll move forward.  But none of the positions proposed here are considered absolute by the project.

David J.

pgsql-hackers by date:

Previous
From: Robert Haas
Date:
Subject: Re: Current master hangs under the debugger after Parallel Seq Scan (Linux, MacOS)
Next
From: Robert Haas
Date:
Subject: Re: making EXPLAIN extensible