3B1 UUCP & Ethernet

[David S. Herron]

In article <34702 at ditka.Chicago.COM> kls at ditka.Chicago.COM (Karl Swartz) writes:
>In article <1991Feb17.061443.3834 at jack.sns.com> steve at jack.sns.com (J. Steven Harrison) writes:
>>There does not seem to be a uucpd available on the system. Does this mean that
>>I cannot do uucp over the ethernet between other machines on the network?
>
>There is a uucpd available as part of the BSD sources on uunet.  This
>is just a front-end which handles the login for incoming connections
>then execs uucico.  It's written for BSD but might work on a 3B1.

uucpd is a Berkeley-ism &so the only uucico's which know how to talk
to uucpd are ones which have been Berkeley-ized.  Suns version of
HDB has the necessary hooks.

>>Is there a public domain NFS version that has been ported to the 3B1?
>
>Nope, and it's unlikely you'll ever see such a thing given the amount
>of necessary kernel support that isn't there.

Yes.. in order to do NFS (correctly) you need support for multiple
file system types in the kernel.  To date the only method used has
been what Sun did (DEC reimplemented the same idea and gave it a
different name).  The basic idea is to insert a layer of code above
the file system specific code & to have a "switch table" to call
file-system specific modules for the file operations.  At the file-system
specific layer the structure is called "inode", and at the layer above
it is called "vnode" (DEC: "gnode").  If you are interested in further
details I'm sure that Bach's book on System V covers this.

In short -- having object files we can link together is no help.
The actual code in there needs to be changed to insert that layer.
If you were to somehow dissassemble (or otherwise obtain source
to) the necessary bits you'd still have to have an NFS.  Well.. I
recall that there was one posted on the net some time back, called
"unfs".  There is also, supposedly, a BSD version of NFS in 4.4BSD.

>>The socket library seems to be rather weak also.
>
>The WIN/3B stuff is very old and creaky.

Yeah..  Veeeeery old.  Like, 5 years old.  (BTW, I'm not in the TCP group
here and nor did I have anything to do with WIN/3B for 3b1's.  The 3b1
it was developed on *is* in my office, though, sitting on top of a
bookshelf .. Finally, current WIN/TCP is very good stuff.. ;-) )

> Alex Crain ported the BSD
>4.3 socket code and had talked about doing SLIP and maybe even support
>for the Ethernet boards but I haven't heard anything more.  (The uipc
>stuff just posted to comp.sources.3b1 is Alex's work and more; there
>isn't a README file but it appears to be close to the version placed
>on osu-cis last year and doesn't have the networking support.)

The uipc stuff is a very good start.  I glanced over it and it
looks pretty good, but is lacking all of the AF_INET code.  I didn't
glance long enough to get an idea of how long it would take to put
the rest of that code in there ... It *would* be a fun project to
port the current TCP/IP.  The other night I estimated it might take
a month of part time work.  I simply don't have anywhere near that
kind of time (I'm working overtime as it is), and even if I did
some conflict of interest and/or support issues might raise their
head.  I *am* willing to provide moral support and encouragement if
someone else will do the work ;-).

[Previous paragraph is because Thad mentioned my name in another posting.]



-- 
<- David Herron, an MMDF & WIN/MHS guy, <david at twg.com>
<- Formerly: David Herron -- NonResident E-Mail Hack <david at ms.uky.edu>
<-
<-	MS-DOS ... The ultimate computer virus.