Please note that, as may be gleaned by my From: and Reply-To:, I do 
not wish to receive a reply via email. Reply only to the list please 
and thanks.

> First, re. the Reply-To: field: This field is TOTALLY invisible to those
> of us who read the list via, e.g., mail-index.netbsd.org.  Please use the
> desired reply address in your From: line or point it out in the body of
> the email (and avoid allowing your mailer to use MIM64 or Quoted-Printable
> or other mangling encoding schemes).

That's unfortunate. That means you're missing the In-reply-to: which 
seems to break many different mailers' threading capabilities. :(

> In addition to replying to "netbsdMLpostNO@SPAM.quake.ca" that
> you list in the From: field, I'll CC this to "netbsdMLpost@quake.ca"
> and hope that you get at least one of those---or that you read this
> list one way or another.

As I stated in my note, I only wish replies to be sent to the mailing 
list. Also I am subscribed to most of them (and read them often.)

> Second, Marc, I suggest that you ask again on either netbsd-help
> or tech-pkg.  (Or, perhaps, current-users, or a port-specific list
> such as port-amd64.)

I forgot a uname -a:

NetBSD draco 2.99.11 NetBSD 2.99.11 (DRACO) #0: Tue Dec 28 09:44:41 
PST 2004  root@draco:/v3/src/sys/arch/i386/compile/DRACO i386

I thought this list would be more appropriate, since I'm running 
-current and the issue is possibly kernel- or core libraries-based. 
port-i386 might be more narrowly appropriate, but I doubt this issue 
is port-related, since to get sound working in SDL to begin with, I 
seem to recall the problem was fixed with semaphores.

Also, upon reflection, xmms and other standard application software 
has a slight delay in it which can sometimes become crackly or broken 
sound on Duron-based machines if the machine is under load. The only 
recovery from such a problem is a restart of the mp3 or sound player.

> particular relation of your issue to -current, that I can see, and
> the group of people on this list are not particularly likely to know
> the answers.

...perhaps tech-kern. Hrm..