On Tue, 15 Apr 1997, Bill Studenmund wrote:

> > 
> > Hello!
> > 
> > I wonder if it is possible to run CAP or netatalk over the printerport
> > under NetBSD.
> 
> No. Neither PPP nor LocalTalk are supported at this time. I'm working on
> LocalTalk, but it'll take a while as I'm also trying to finish my PhD. :-)
> 
> The actual drivers aren't hard. The hard part is making a clean design
> which can share the tty with normal serial port uses (all the Linux
> LocalTalk drivers I've seen are for cards which only support LocalTalk
> so they don't have the sharing problem).

Would it be less difficult to create a version which just blows off the
"normal serial port uses"?  I, for one, would consider a LocalTalk Mac-to-
NetBSD connection to be nothing short of miraculous, and would give up a
certain amount of functionality to obtain it.

Maybe I misunderstand what you mean? Do you mean "normal serial port uses"
only on the port devoted to AppleTalk, or on both ports? (I would love to
be able to dial into my NetBSD box and access the files on my PBase...)

_That_ would merit a triple :-)   !!!!



> 
> > As far as I know, netatalk has to build during kernel compilation, but
> > runs as a demon, right?
> > I'm unable to compile a kernel, so i have to use CAP.
> 
> Yes. Netatalk puts ddp into the kernel. ddp is the closest Appletalk
> equivelent to IP. It sends packets from one machine to the next with no
> assurance of delivery. All the other layers are in programs, and the
> atalkd daemon sits around and helps run things.
> 
> > If it is possible to run CAP over the printerport, i would like to
> > build a LocalTalk-net with my Q700 running NetBSD and my 165c running
> > MacOS.
> > 
> > And if this is possible, will it be possible to mount the hfs-volumes
> > that are on the scsi-bus of the BSD-Mac on the 165c Powerbook?
> > Like this:
> > hfs-volume(SCSI-bus) -> Q700(NetBSD)
> >  -> CAP(Printerport) -> PB 165c(Icon on desktop from hfs-volume)
> 
> Not really. Paul Goyette and I worked on supporting this, but never
> finished. We got an hfs file system going but never cured a bug in its
> buffer management. We never figured it out.
> 
> I'd think that a better thing would be to start over with hfs_fs. It's
> VERY Linuxy, but it seems to be writing ok and its author is iterested
> in a NetBSD port. :-)
> 

Might there be a vague time-frame for a minimally useful version (see
above...)? (We won't hold you to it, Bill...)


J.