At 10:44 28.09.99 -0700, Jason Thorpe wrote:
>On Tue, 28 Sep 1999 10:29:37 +0200
>  Hauke Fath <hf@Melog.DE> wrote:
>
>I have a load of changes I'll be committing very soon, with a lot more
>to come.

That's good to hear, Jason.

Do you think you might find the time to provide documentation? Apart from 
quality issues (again, compare the {Net,Open}BSD st.c 
<http://www.freebsd.org/cgi/cvsweb.cgi/syssrc/sys/dev/scsipi/st.c?cvsroot=ne 
tbsd>
to FreeBSD's scsi_sa.c 
<http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/cam/scsi/scsi_sa.c?cvsroot=fr 
eebsd> ; compare the comments, and, especially, compare the commit logs), I 
feel that lack of documentation for the architecture is one of the biggest 
flaws of the "old" *BSD scsi subsystem.

If you were sitting on i386 (I am a mac68k person) and you were looking at 
SCSI issues and an OS to work on, would you choose FreeBSD or NetBSD? I 
have always seen NetBSD's scsi subsystem as a twisted maze.

>We will never have the FreeBSD CAM code.  Its device model doesn't fit
>into NetBSD's autoconfiguration scheme.  We will have some of the features
>it has, though.

A pity. Can you elaborate on "doesn't fit"?

Two major points to gain would be documentation on architecture
( <ftp://ftp.t10.org/t10/drafts/cam/cam-r12b.pdf> ,
<ftp://ftp.t10.org/t10/drafts/cam3/cam3r03.pdf> , and DEC/Compaq has a lot 
of stuff, too) and driver support from FreeBSD (where is AIC789x support?).

         hauke

--
Hauke Fath                              Saphir Software GmbH
                                         D-69115 Heidelberg
hf@SaphirSC.DE                          Ruf +49-6221-13866-35, Fax -21