Xref: utzoo comp.protocols.nfs:1941 comp.windows.ms:10213
Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!samsung!munnari.oz.au!bruce!trlluna!titan!craick
From: craick@titan.trl.oz (John Craick)
Newsgroups: comp.protocols.nfs,comp.windows.ms
Subject: Re: Sun PC-NFS deficiencies
Message-ID: <2799@trlluna.trl.oz>
Date: 13 Mar 91 03:21:32 GMT
References: <1991Mar7.185009.27239@amd.com> <1991Mar9.120940.23851@ukpoit.co.uk> <1991Mar11.232450.5556@amd.com>
Sender: news@trlluna.trl.oz
Lines: 61



The debate about PC-NFS "deficiencies" is interesting in several
respects and it seems there may well be an outbreak of tribal warfare on
the issue. I wonder, incidentally, why the Windows/PC-NFS problem is
seen as "belonging" to Sun as opposed to Microsoft.

With respect, I'd like to reinforce someone's previous point that one of
the most important questions is actually what sort of network services &
facilities will best meet the needs of users given their particular
tasks. Is a Unix network, a PC network or some combination of these the
best choice ? Is the user, and, indeed, their organization when it comes
down to it, best served by Unix services and applications or by PC
s&a's ? PC-NFS is not perfect, of course, but it does allow powerful
mixed combinations to be used.

Just for the record, using PC-NFS 3.0 or thereabouts, plus 386 class
PC's with plenty of memory and without any configuration switching, you
CAN have :

(1) MS Windows in 386 enhanced mode
(2) Telnet sessions run on Unix hosts, inside or outside Windows,  with at
    least VT 100 terminal capabilities - i.e. Vi to your hearts content.
(3) FTP sessions from DOS, inside Windows or on a Unix host
(4) Network mounted DOS disks on which DOS application data can be
    readily created, accessed and stored and from which many (tho' not all)
    DOS applications can actually be run.
(5) Network mounted DOS printers (LPT2 ...) - but see below
(6) 585 Kbyte usable memory space under DOS
(7) About 560 Kbyte usable memory in  DOS windows under Windows.

To me, all this seems a pretty powerful combination but there are still some
problems areas.

Firstly, printing. So far, tho' netmounted (postcript) LPTs work fine
under DOS, I haven't persuaded Windows to take any notice of them - a
considerable nuisance but work-roundable (!). Maybe I haven't found the
right recipe yet, tho' others suggest they have. Also, since all this is
a "no network" as far as WINDOWS is concerned, there's no direct means
of inspecting or manipulating the network print queue. You can, however,
telnet access the Unix print host and operate to some extent on the
print queue there.

Secondly, at present in my environment, multitasking & task switching,
normally available under Windows, seem a little wobbly when network
operations are involved. Maybe I haven't found the right recipes and
settings or maybe there are some basic problems. It would be nice if the
new PC-NFS bits came with some .pif files.

Overall, there isn't yet any single, ideal and perfect operating/working
environment and there probably won't ever be. Even the "old" PC-NFS was
quite a useful tool and things seem to be evolving in several useful
ways. Finally, I'd suggest that, while critical debate is  probably
useful, tribal warfare and dismissal of this, that or the other style of
network or this or that company or package isn't likely to help anyone
very much.


John Craick

(j.craick@trl.oz.au)