Good news and bad news about the 1.6Q snapshot announced to this list
the other day.

Good News:

It works! Mostly, anyway. 

I swapped disks (to preserve my "production" setup) in my SS20 (2xSM81,
512MB RAM) and installed 1.6Q from the snapshot. There were some
"oddities" on the install, though. First, I tried installing (_not_
updating) with "use existing" since the "spare" disk I swapped in
already contained a NetBSD disklabel. The install seemed to go OK but
simply overwrote existing files with new files like it was doing an
upgrade instead of doing a newfs and wiping the partitions. I wasn't
happy with that so I went back and did an install again, this time
partitioning the disk. The installer detected existing partitions, sort
of, but with bizarre begin and end points. No setback, since I was
repartitioning the disk anyway, but it sure seemed weird.

The install went mostly OK but, unlike previous installs I've done, the
network information was NOT saved to the target disk and I had to
manually configure the network after reboot. Further, although an MP
kernel set was present, the installer did not offer to install it for
me. I just went ahead and accepted the GENERIC and manually installed
the MP kernel afterward.

MP came up in multi-user and the system happily accepted about half a
dozen telnet sessions (after I modified inetd.conf, of course) without a
hitch. Performance seemed very good - better than The Other OS I usually
run on this box (you'll see why I use it below). I installed some
software via pkgsrc and would have been very happy except for one thing.

Bad News:

Parallel printing _doesn't_ work.

I installed CUPS 1.1.18 via pkgsrc, but never got a test page printed.
Part of the page would print OK but then it just started printing
garbage (and wasting lots of paper with random page ejects). The printer
I was printing to is an Epson C80 which works very well with the
hardware when running Solaris 9. In fact, this is my "production" home
print server setup.

To even get CUPS to recognize there _was_ a parallel port, I had to
create a symlink from //dev/bpp to /dev/lpt0 or /dev/lpa0. I only
figured this out after running 'strings' on the parallel backend. Using
/dev/lpt0 tells the parallel backend to use interrupts while using
/dev/lpa0 says to use polling. I got the same results either way.

No News:

I didn't try anything under X.


Just a data point for those interested :-)

Cheers!

-- 
Bill Meahan  WA8TZG  wmeahan@wa8tzg.org  /~\  The ASCII Ribbon Campaign
"Never ascribe to mailce what can be     \ /    No HTML/RTF in email
   adequately explained by stupidity      X     No Word docs in email
   -- Robert Heinlein                    / \  Respect for open standards