configure --with-uuid=bsd fails on NetBSD

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: configure --with-uuid=bsd fails on NetBSD
Date: 2022-08-20 23:39:32
Message-ID: 3848059.1661038772@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Our documentation claims that --with-uuid=bsd works on both
FreeBSD and NetBSD: installation.sgml says

<option>bsd</option> to use the UUID functions found in FreeBSD, NetBSD,
and some other BSD-derived systems

and there is comparable wording in uuid-ossp.sgml.

In the course of setting up a NetBSD buildfarm animal, I discovered
that this is a lie. NetBSD indeed has the same uuid_create() function
as FreeBSD, but it produces version-4 UUIDs not version-1, which causes
the contrib/uuid-ossp regression tests to fail. You have to dig down
a level to the respective uuidgen(2) man pages to find documentation
about this, but each system appears to be conforming to its docs,
and the old DCE standard they both refer to conveniently omits saying
anything about what kind of UUID you get. So this isn't a bug as
far as either BSD is concerned.

I'm not personally inclined to do anything about this; I'm certainly
not excited enough about it to write our own v1-UUID creation code.
Perhaps we should just document that on NetBSD, uuid_generate_v1()
and uuid_generate_v1mc() don't conform to spec.

regards, tom lane

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Andres Freund 2022-08-21 00:42:34 Re: [RFC] building postgres with meson
Previous Message Peter Geoghegan 2022-08-20 23:28:59 Instrumented pages/tuples frozen in autovacuum's server log out (and VACUUM VERBOSE)