DNA

[Richard Lamson <rsl@max-fleischer.ila-sf.dialnet.symbolics.com>]

    Date: Tue, 29 Nov 88 09:48 EST
    From: Eric S. Crawley <Crawley@ALDERAAN.SCRC.Symbolics.COM>

	Date: Mon, 28 Nov 88 21:07 PST
	From: York@Chuck-Jones.ILA-West.Dialnet.Symbolics.COM (William M. York)

	    Date: Mon, 28 Nov  14:45:07 1988
	    From: whuts!milich@att.att.com

	    Does anyone know of an alternative to DNA between Symbolics Genera 7.2
	    and VAX VMS 4.4? We would like to use the VAX as a file server that is
	    capable of handling the Load/Compile system feature of Symbolics.
	    In addition we would like to store documentation files on the VAX
	    and use the Symbolics document examiner. In other words we want to do
	    more than just file transfers - we would like the VAX to do what a 
	    Symbolics "file server" would do. Any suggestions or leads are greatly
	    appreciated. Thank you in advance.

	Symbolics now has an NFS network protocol implementation on their price
	list.  You might be able to use it in conjunction with a VMS NFS server
	to get the results you want.  I have included Richard Lamson in the
	distribution of this message 'cause I think that he has recent
	experience with some NFS package for VMS.

The NFS package for VMS that I know about is from Wollongong, and is not
very well adapted to use from anything but Unix.  They translate all the
"complicated" VMS pathname features into something else.  For example,
if your host knows that their host supports version numbers and insists
on sending them in the file name strings, the file names on the VMS side
become "FOO$092BAR$094103.RANDOM;1" instead of "FOO.BAR;103" (I don't
remember the actual translations, but you get the idea.)  Thus, the VMS
user doesn't get to see true VMS pathnames.  

The problem, of course, is that NFS as currently implemented is too
Unix-specific (or rather, they didn't bother to make it generic enough,
since, after all, who would want to use anything other than Unix?  What
would you do with version numbers other than get into trouble?)  Chris
Lindblad suggested a complicated scheme for fixing this problem, but I
would be very surprised if Wollongong ever changes anything unless Sun
changes the protocol.

    Another alternative, if you can only buy software for your VAX, is a
    package for VMS VAXen called MultiNet that allows a VAX to talk IP/TCP
    and Chaos.  The chaos QFILE protocol is supported in MultiNet so the
    Symbolics machines can use the VAX as a real file server without any
    additional software on the Symbolics machines.  MultiNet was developed
    at SRI but, last I heard, the developers had left to form their own
    company to support MultiNet.  The catch is that it is rather expensive
    (on the order of $10k per VAX CPU).  The pricing may have changed
    recently.

QFILE also has a number of deficiencies, which is why NFile was
invented.  You are unlikely to see any of these except the "Untimely
Abort" problem; keep your cotton-pickin' fingers off the control-abort
key if you use it.

I guess I don't understand what's wrong with DNA.  It supports version
numbers, more-or-less works on the Symbolics side, and is well supported
by DEC.  I don't konw how much support it receives on the Symbolics
side, but perhaps it is stable enough that it doesn't really need any.