Re: [PATCH] random_normal function - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: [PATCH] random_normal function
Date
Msg-id aONFybtRwxU2v9b2@paquier.xyz
Whole thread Raw
In response to Re: [PATCH] random_normal function  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On Sat, Oct 04, 2025 at 02:07:45PM -0400, Tom Lane wrote:
> Actually, there is a *ton* of obsolete information on that page:
> there's no serial_schedule file anymore, and we don't have
> dynamically-generated test files anymore (input/output subdirectories)
> and therefore there's no substituted variables.  The age of the whole
> thing can be dated by its references to CVS :-(
>
> Maybe we should indeed try to replace it with something in the main
> docs, which are a bit more likely to get maintained.  There is an
> existing chapter "Regression Tests", but it presently only talks
> about how to run the tests not how to write new ones.

Yes, most of the content of this page could be deleted.

IMO, the main docs would benefit greatly from the addition of
PG_ABS_BUILDDIR, PG_ABS_BUILDDIR, PG_LIBDIR and PG_DLSUFFIX, which are
still missing.  These are available to extension developers through
installcheck, and are a great way to write more flexible tests with
arbitrary inputs, like a dependency to on-disk files.  I would
certainly +1 this addition, at least.
--
Michael

Attachment

pgsql-hackers by date:

Previous
From: Michael Paquier
Date:
Subject: Executing pg_createsubscriber with a non-compatible control file
Next
From: David Rowley
Date:
Subject: Re: Should we update the random_page_cost default value?