[TUHS] networking on unix before uucp

[TUHS] networking on unix before uucp

[Mark Longridge]

Hi folks,

I was wondering if Unix had any form of networking before uucp
appeared in Unix v7. It would be interesting to know if one could pass
a file from one Unix v5 machine to another without having to store it
on a magnetic tape.

There's some reference to a mysterious "Spider Interface" in the Unix
v5 manual. It seems to have something to do with DR-11B (which is a
general purpose direct memory access interface to the PDP-11 Unibus).

There's also reference to the "Spider line-printer" :)

Mark

[TUHS] networking on unix before uucp

[Mary Ann Horton]

At Bell Labs, PWB (1, I think, which was V6-based) had a sophisticated 
RJE engine that could submit jobs to the GE mainframe and get the output 
back (hence the GECOS field in /etc/passwd.) This was also used to 
submit jobs to be printed using the opr (off-line print) command.

At Berkeley, in 1978 we originally had a "cross-over cable" (basically a 
null modem, my lousy soldering job) in our patch panel, to allow two 
UNIX boxes to cat files across - not even Kermit, as I recall.

Shortly thereafter in 1978, Eric Schmidt (yes, that Eric Schmidt) wrote 
Berknet, which was similar to UUCP but didn't use modems, it ran over 
null modem serial line interconnections.  It ran on V6, V7, and the Vax.

For a few years, the Berknet link between ucbvax (which had a modem and 
was on UUCP) and ingvax (which was on the ARPANET) was the gateway 
between the UUCP and Usenet networks and the ARPANET.

On 08/21/2014 08:42 PM, Mark Longridge wrote:
> Hi folks,
>
> I was wondering if Unix had any form of networking before uucp
> appeared in Unix v7. It would be interesting to know if one could pass
> a file from one Unix v5 machine to another without having to store it
> on a magnetic tape.
>
> There's some reference to a mysterious "Spider Interface" in the Unix
> v5 manual. It seems to have something to do with DR-11B (which is a
> general purpose direct memory access interface to the PDP-11 Unibus).
>
> There's also reference to the "Spider line-printer" :)
>
> Mark
> _______________________________________________
> TUHS mailing list
> TUHS at minnie.tuhs.org
> https://minnie.tuhs.org/mailman/listinfo/tuhs

[TUHS] networking on unix before uucp

[Cory Smelosky]

On Fri, 22 Aug 2014, Mary Ann Horton wrote:

>
[snip]
> Shortly thereafter in 1978, Eric Schmidt (yes, that Eric Schmidt) wrote 
> Berknet, which was similar to UUCP but didn't use modems, it ran over null 
> modem serial line interconnections.  It ran on V6, V7, and the Vax.
>

How does that Berknet differ from the Berknet in...4.0/4.1BSD?  I've seen 
what I THINK was ethernet code there...but I couldn't be sure as I 
couldn't even decipher the addressing scheme. ;)

I've been meaning to ask about Berknet, anyway.  One of my side projects 
is to get it operational.

> For a few years, the Berknet link between ucbvax (which had a modem and was 
> on UUCP) and ingvax (which was on the ARPANET) was the gateway between the 
> UUCP and Usenet networks and the ARPANET.
>

-- 
Cory Smelosky
http://gewt.net Personal stuff
http://gimme-sympathy.org Projects

[TUHS] networking on unix before uucp

[Clem Cole]

4.1 did not have Ethernet support from UCB. only Berknet.   the original ip stack with Ethernet support for 4.1 was done at BBN.   the interface is classic unix using the open call.    similar to the MIT ChaosNet stack

CMU had a new os for the TripleDrip PERQ called Accent. it had a number of interesting concepts such as ports.    

Joy took the BBN stack and created Berkeley sockets as a reaction to Accent's networking scheme.  this would become 4.1A/B/C and eventually 4.2




btw.   CMU responded to 4.2 by taking the ideas from Accent and rewriting then and splicing them into BSD kernel to create Mach.      Which lives today as the core of both Mac OSx and iOS

Clem

> On Aug 22, 2014, at 6:12 PM, Cory Smelosky <csmelosky at gewt.net> wrote:
> 
>> On Fri, 22 Aug 2014, Mary Ann Horton wrote:
>> 
> [snip]
>> Shortly thereafter in 1978, Eric Schmidt (yes, that Eric Schmidt) wrote Berknet, which was similar to UUCP but didn't use modems, it ran over null modem serial line interconnections.  It ran on V6, V7, and the Vax.
> 
> How does that Berknet differ from the Berknet in...4.0/4.1BSD?  I've seen what I THINK was ethernet code there...but I couldn't be sure as I couldn't even decipher the addressing scheme. ;)
> 
> I've been meaning to ask about Berknet, anyway.  One of my side projects is to get it operational.
> 
>> For a few years, the Berknet link between ucbvax (which had a modem and was on UUCP) and ingvax (which was on the ARPANET) was the gateway between the UUCP and Usenet networks and the ARPANET.
> 
> -- 
> Cory Smelosky
> http://gewt.net Personal stuff
> http://gimme-sympathy.org Projects
> _______________________________________________
> TUHS mailing list
> TUHS at minnie.tuhs.org
> https://minnie.tuhs.org/mailman/listinfo/tuhs

[TUHS] networking on unix before uucp

[Clem Cole]

btw: the addressing scheme in Berknet is pretty simple.  it's mupltiplexing an 9600 baud rs232c connection point to point without full connectivity (like uucp).   although IIRC the sender did not have to specify the path.   Berknet figured it out for you - but I might be confusing a different network scheme from those days.  there were so many

Cl

> On Aug 24, 2014, at 6:46 PM, Clem Cole <clemc at ccc.com> wrote:
> 
> 4.1 did not have Ethernet support from UCB. only Berknet.   the original ip stack with Ethernet support for 4.1 was done at BBN.   the interface is classic unix using the open call.    similar to the MIT ChaosNet stack
> 
> CMU had a new os for the TripleDrip PERQ called Accent. it had a number of interesting concepts such as ports.    
> 
> Joy took the BBN stack and created Berkeley sockets as a reaction to Accent's networking scheme.  this would become 4.1A/B/C and eventually 4.2
> 
> 
> 
> 
> btw.   CMU responded to 4.2 by taking the ideas from Accent and rewriting then and splicing them into BSD kernel to create Mach.      Which lives today as the core of both Mac OSx and iOS
> 
> Clem
> 
>>> On Aug 22, 2014, at 6:12 PM, Cory Smelosky <csmelosky at gewt.net> wrote:
>>> 
>>> On Fri, 22 Aug 2014, Mary Ann Horton wrote:
>> [snip]
>>> Shortly thereafter in 1978, Eric Schmidt (yes, that Eric Schmidt) wrote Berknet, which was similar to UUCP but didn't use modems, it ran over null modem serial line interconnections.  It ran on V6, V7, and the Vax.
>> 
>> How does that Berknet differ from the Berknet in...4.0/4.1BSD?  I've seen what I THINK was ethernet code there...but I couldn't be sure as I couldn't even decipher the addressing scheme. ;)
>> 
>> I've been meaning to ask about Berknet, anyway.  One of my side projects is to get it operational.
>> 
>>> For a few years, the Berknet link between ucbvax (which had a modem and was on UUCP) and ingvax (which was on the ARPANET) was the gateway between the UUCP and Usenet networks and the ARPANET.
>> 
>> -- 
>> Cory Smelosky
>> http://gewt.net Personal stuff
>> http://gimme-sympathy.org Projects
>> _______________________________________________
>> TUHS mailing list
>> TUHS at minnie.tuhs.org
>> https://minnie.tuhs.org/mailman/listinfo/tuhs

[TUHS] networking on unix before uucp

[A. P. Garcia]

On Aug 24, 2014 5:47 PM, "Clem Cole" <clemc at ccc.com> wrote:

> btw.   CMU responded to 4.2 by taking the ideas from Accent and rewriting
then and splicing them into BSD kernel to create Mach.      Which lives
today as the core of both Mac OSx and iOS

And of course, before that, nextstep/openstep... and, well, hurd.  ;-)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140824/10895399/attachment.html>

[TUHS] networking on unix before uucp

[Clem Cole]

next used Mach.  you're right next brought it into Apple

> On Aug 24, 2014, at 7:07 PM, "A. P. Garcia" <a.phillip.garcia at gmail.com> wrote:
> 
> 
> On Aug 24, 2014 5:47 PM, "Clem Cole" <clemc at ccc.com> wrote:
> 
> > btw.   CMU responded to 4.2 by taking the ideas from Accent and rewriting then and splicing them into BSD kernel to create Mach.      Which lives today as the core of both Mac OSx and iOS
> 
> And of course, before that, nextstep/openstep... and, well, hurd.  ;-)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140824/4c7fde61/attachment.html>

[TUHS] networking on unix before uucp

[Ronald Natalie]

On Aug 24, 2014, at 7:07 PM, A. P. Garcia <a.phillip.garcia at gmail.com> wrote:

> 
> On Aug 24, 2014 5:47 PM, "Clem Cole" <clemc at ccc.com> wrote:
> 
> > btw.   CMU responded to 4.2 by taking the ideas from Accent and rewriting then and splicing them into BSD kernel to create Mach.      Which lives today as the core of both Mac OSx and iOS
> 
> And of course, before that, nextstep/openstep... and, well, hurd.  ;-)
> 
> 
Amusingly, when we got our first NeXT machines I was just poking around working on it and inadvertently typed "bg" to /bin/sh.

I got back "Job Control not Enabled."

Hey, that error message sounds familiar.   So I typed "set -J" 

"Job Control Enabled."

Hey, this is one my shells.    It has made it to NeXt via Mach via Doug Gwyn's System V on BSD tapes.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140825/95a4d593/attachment.html>

[TUHS] networking on unix before uucp

[Jeremy C. Reed]

On Fri, 22 Aug 2014, Cory Smelosky wrote:

> How does that Berknet differ from the Berknet in...4.0/4.1BSD?  I've 
> seen what I THINK was ethernet code there...but I couldn't be sure as 
> I couldn't even decipher the addressing scheme. ;)
>
> I've been meaning to ask about Berknet, anyway.  One of my side 
> projects is to get it operational.

I have read most (if not all) of the berknet docs and a lot of the code. 
The docs and code referenced a rcs, rcsq, and rcslog as some tools that 
could be ran without a personal account on a remote system. I didn't 
recognize rcsq, but also these references pre-dated Tichy's alternative 
to SCCS by a year or so.

Then I realized this is "remote computer system link" used something 
like: rcsq to see queued jobs no sent yet for CDC 6400; rcslog to see 
history of jobs sent that day; and rcsrm to delete a job not sent yet. 
Basic concepts similar to berknet (netq, netlog, netrm).
(The clue I found was in 
http://www.eecs.berkeley.edu/Pubs/TechRpts/1979/ERL-79-16.pdf)

I know the Cyber 6400 was used there (some of my interviewees told me 
about it). Does anyone know if Berknet's design was inspired by this 
"rcs"?  Where can I learn more about it? (as searching for "rcs" is 
difficult)

I don't see any ethernet code in Berknet. The author had a summer job at 
XEROX PARC I think in the middle of his Berknet project and discussed 
the low-level network concepts with Boggs. Later, they considered an 
LNI, an early token ring (if I understand correctly), device, and DMC-11 
link, but I don't think berknet was ever extended for those or used 
using ethernet.

Note that extending Berknet probably didn't make sense. It was all 
batched with all the systems known to each system hardcoded, compiled 
in. Smallest jobs sent first from the queue to neighboring system which 
would send the job to next as if it was queued locally until it arrived 
to desired system.  So retrieving an email (box) or doing a remote copy, 
you would have to wait for the jobs to get ran. The speed on the 1200 
baud links was only around 50 characters per second. (I think at best on 
9600 baud links was around 600 characters per second but normally 350 
cps.) In addition, berknet had hardcoded restrictions limiting 100,000 
characters per single job. (It was later extended to 500,000 characters 
for some machines, then all machines while still limiting only 200,000 
characters max size jobs during day time.) (I never used berknet only 
read code and docs and did some interviews with users.) It was quite 
limited compared to the new real-time tools and near 3Mb network that 
started being developed and used there a couple years later.

By the way, the early berknet had a symlink to the batch remote copy 
netcp called "rcp" which predated Joy's real-time remote file copy, rcp, 
by a few years. The early version also included a trivial "sendmail" to 
remotely send mail over the batch berknet to another berknet system (by 
running "mail" on the remote system later). It predated Allman's 
unrelated intelligent mailer by around two years. The Berknet tool was 
replaced and renamed with "sendberkmail".

[TUHS] networking on unix before uucp

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3193 bytes --]

​below​


On Tue, Aug 26, 2014 at 12:36 PM, Jeremy C. Reed <reed at reedmedia.net> wrote:

> On Fri, 22 Aug 2014, Cory Smelosky wrote:
>  Does anyone know if Berknet's design was inspired by this
> "rcs"?  Where can I learn more about it? (as searching for "rcs" is
> ​ ​
> difficult)
>
​You'd have to ask Eric.   It was certainly there and the CDC box was used
a good bit before the Vaxen showed up, particular in EE for things like
SPICE, SPLICE, et al.

>
> I don't see any ethernet code in Berknet.

​There would not be any.   As I said, it pre-dated any Ethernet HW at UCB.
  It ran over 9600 bit RS-232C links (usually on a DZ11).​




> they considered an
> LNI, an early token ring (if I understand correctly), device,

​Yes - that is right.   It was a UCI/MIT device -- predates the Proteon
Ring products and Apollo's Ring.​




> and DMC-11
> link, but I don't think berknet was ever extended for those or used
> using ethernet.
>
​Correct​




>
>
> By the way, the early berknet had a symlink to the batch remote copy
> netcp called "rcp" which predated Joy's real-time remote file copy, rcp,
> by a few years.

​right.​





>  The early version also included a trivial "sendmail" to
> remotely send mail over the batch berknet to another berknet system (by
> running "mail" on the remote system later). It predated Allman's
> unrelated intelligent mailer by around two years. The Berknet tool was
> replaced and renamed with "sendberkmail".


​It was built using Kurt Shoen's "delivermail" which is in 4.1 (and was
what Eric would rewrite - see below).

The historical (hysterical) ​reason was this.

Eric was the main guy behind the 11/70 that the Ingress project had (ing70
on the Berknet) in Cory Hall. In Evans, was Ernie (and later ucbvax).   The
Internet connection was owned by the Ingress Group on a long interface pair
to an IMP at LBL.    UCB did not have its own IMP as CMU, MIT, Stanford
did.   The UUCP network came into Evans via Ernie, but Internet via Ing70.
  So the UCB mail system had to send the messages over the BerkNet to
proper host for import/export from campus.

Similarly, folks in other departments as they joined the Berknet had
private and strange mail interfaces.
Originally, different folks in different departments had been hacking mail,
delivermail etc al to handle different header formats.   Eric was a DB guy,
do he wrote a DB production language to walk headers because it was having
to deal with so much dead mail from unparsable messages that would get sent
to Ingress by the Berknet and then not be able to be delivered.

I have always contended that if he has left the SMTP connection out of
sendmail and called the BBN smtpd, to do import/export the way many other
UNIX MTAs did and left sendmail and as purely a middleware layer that
canonnicalized header, it would have been a different world.    Because it
was also the smtpd, people used it ever though most did not have the N
header problem we had at UCB.


The rest is history.
Clem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140826/92c959be/attachment.html>

[TUHS] networking on unix before uucp

[Dan Cross]

Unix was on the ARPAnet circa 1975 (if not earlier):
http://tools.ietf.org/html/rfc681

V6 was "released" in May 1975, and that document was also published in May
1975 and says that the software had been running for about a month, so it's
entirely possible that the ARPAnet Unix work was done before V6 (or perhaps
they were an early test site: I don't know what the policies were around
that).  It's been many years since I've read RFC681 closely and from my
quick skim just now, I don't think they say what version of Unix they're
running.  It's clear from the RFC that they had been running Unix for more
than a month given the description of their site, and if I had to hazard a
guess I'd say they were running V5; perhaps heavily patched.  I idly wonder
if any of that work has survived; it would be interesting to see an
ARPAnet/NCP implementation for early Unix.

But to address your question yes, Unix was certainly networked well before
UUCP emerged in V7.

        - Dan C.



On Thu, Aug 21, 2014 at 11:42 PM, Mark Longridge <cubexyz at gmail.com> wrote:

> Hi folks,
>
> I was wondering if Unix had any form of networking before uucp
> appeared in Unix v7. It would be interesting to know if one could pass
> a file from one Unix v5 machine to another without having to store it
> on a magnetic tape.
>
> There's some reference to a mysterious "Spider Interface" in the Unix
> v5 manual. It seems to have something to do with DR-11B (which is a
> general purpose direct memory access interface to the PDP-11 Unibus).
>
> There's also reference to the "Spider line-printer" :)
>
> Mark
> _______________________________________________
> TUHS mailing list
> TUHS at minnie.tuhs.org
> https://minnie.tuhs.org/mailman/listinfo/tuhs
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140822/30d86164/attachment.html>

[TUHS] networking on unix before uucp

[Dan Cross]

(And upon slightly closer inspection of the RFC's header, it appears that
it was *written* in March of 1975, though Postel didn't post it until May
of that year.  That certainly predates the V6 release by a few months, so
it seems probable that they were, in fact, running either V5 [possibly with
patches] or a pre-release of V6.)


On Fri, Aug 22, 2014 at 11:18 AM, Dan Cross <crossd at gmail.com> wrote:

> Unix was on the ARPAnet circa 1975 (if not earlier):
> http://tools.ietf.org/html/rfc681
>
> V6 was "released" in May 1975, and that document was also published in May
> 1975 and says that the software had been running for about a month, so it's
> entirely possible that the ARPAnet Unix work was done before V6 (or perhaps
> they were an early test site: I don't know what the policies were around
> that).  It's been many years since I've read RFC681 closely and from my
> quick skim just now, I don't think they say what version of Unix they're
> running.  It's clear from the RFC that they had been running Unix for more
> than a month given the description of their site, and if I had to hazard a
> guess I'd say they were running V5; perhaps heavily patched.  I idly wonder
> if any of that work has survived; it would be interesting to see an
> ARPAnet/NCP implementation for early Unix.
>
> But to address your question yes, Unix was certainly networked well before
> UUCP emerged in V7.
>
>         - Dan C.
>
>
>
> On Thu, Aug 21, 2014 at 11:42 PM, Mark Longridge <cubexyz at gmail.com>
> wrote:
>
>> Hi folks,
>>
>> I was wondering if Unix had any form of networking before uucp
>> appeared in Unix v7. It would be interesting to know if one could pass
>> a file from one Unix v5 machine to another without having to store it
>> on a magnetic tape.
>>
>> There's some reference to a mysterious "Spider Interface" in the Unix
>> v5 manual. It seems to have something to do with DR-11B (which is a
>> general purpose direct memory access interface to the PDP-11 Unibus).
>>
>> There's also reference to the "Spider line-printer" :)
>>
>> Mark
>> _______________________________________________
>> TUHS mailing list
>> TUHS at minnie.tuhs.org
>> https://minnie.tuhs.org/mailman/listinfo/tuhs
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140822/5252b9cc/attachment.html>

[TUHS] networking on unix before uucp

[John Cowan]

Dan Cross scripsit:

> Unix was on the ARPAnet circa 1975 (if not earlier):
> http://tools.ietf.org/html/rfc681

#     filedes = open( "/dev/net/harv",2 );
#     if( filedes < 0 )
#        printf(" harvard is dead");
#     else
#        while( (nbytes=read(filedes,buf,80)) > 0 )
#           write( 0,buf,nbytes );                                   

If only this code still worked on modern Unixes!  The socket API is
fine, but there really was no need to break good old open, at least
for client-side operations.  Plan 9 got it right here, as usual.

#  In this light Bell was approached to see what their reaction
#  would be to an ARPA network wide liscense,  they  said  they were
#  open  to suggestions in that area.  So should enough people
#  become interested, perhaps a less expensive fee can be
#  negotiated.                                                        

Alas for maiden, alas for Judge,
For rich repiner and household drudge!
God pity them both! and pity us all,
Who vainly the dreams of youth recall;
For of all sad words of tongue or pen,
The saddest are these: "It might have been!"
    --John Greenleaf Whittier, "Maud Muller"

To which Bret Harte added in "Mrs. Judge Jenkins":

More sad are these we daily see:
"It is, but hadn't ought to be".

-- 
John Cowan          http://www.ccil.org/~cowan        cowan at ccil.org
The whole of Gaul is quartered into three halves.
        --Julius Caesar

[TUHS] networking on unix before uucp

[Ronald Natalie]

I hate the socket interface, at least once you use it's goofy interface on UNIX, it works mostly like a file descriptor.
The sucky one is Windows which has a socket interface but it's file read and write calls are completely incompatible.

[TUHS] networking on unix before uucp

[Dave Horsfall]

On Fri, 22 Aug 2014, Ronald Natalie wrote:

> I hate the socket interface, at least once you use it's goofy interface 
> on UNIX, it works mostly like a file descriptor.

In a previous slavery we had a simple socket library; the application did 
little more than say whether it was a client or a server etc.  I wish I 
could steal that code now...

> The sucky one is Windows which has a socket interface but it's file read 
> and write calls are completely incompatible.

Well, we mustn't imitate Unix, must we?

-- Dave

[TUHS] networking on unix before uucp

[Larry McVoy]

On Sat, Aug 23, 2014 at 04:11:42AM +1000, Dave Horsfall wrote:
> On Fri, 22 Aug 2014, Ronald Natalie wrote:
> 
> > I hate the socket interface, at least once you use it's goofy interface 
> > on UNIX, it works mostly like a file descriptor.
> 
> In a previous slavery we had a simple socket library; the application did 
> little more than say whether it was a client or a server etc.  I wish I 
> could steal that code now...

lmbench has one:

/*
 * Get a TCP socket, bind it, figure out the port,
 * and advertise the port as program "prog".
 *
 * XXX - it would be nice if you could advertise ascii strings.
 */
int
tcp_server(int prog, int rdwr)

/*
 * Unadvertise the socket
 */
int
tcp_done(int prog)

/*
 * Accept a connection and return it
 */
int
tcp_accept(int sock, int rdwr)

/*
 * Connect to the TCP socket advertised as "prog" on "host" and
 * return the connected socket.
 *
 * Hacked Thu Oct 27 1994 to cache pmap_getport calls.  This saves
 * about 4000 usecs in loopback lat_connect calls.  I suppose we
 * should time gethostbyname() & pmap_getprot(), huh?
 */
int
tcp_connect(char *host, int prog, int rdwr)

void
sock_optimize(int sock, int flags)

int
sockport(int s)

I swiped all that code and we use it in bitkeeper.

        sock = tcp_server(0, port, 0);
        if (sock == -1) exit(1);
        verbose((stderr, "started server on port %d\n", sockport(sock)));

        while (1) {
                if ((nsock = tcp_accept(sock)) < 0) continue;
                peer = peeraddr(nsock);
                verbose((stderr, "connection from %s\n", peer));
                fin = fdopen(nsock, "r");
                fout = fdopen(nsock, "w");
                info_cmds(fin, fout, dashx);
                fclose(fin);
                fclose(fout);
                verbose((stderr, "%s is done\n", peer));
        }

-- 
---
Larry McVoy            	     lm at mcvoy.com             http://www.mcvoy.com/lm 

[TUHS] networking on unix before uucp

[John Cowan]

Larry McVoy scripsit:

> lmbench has one:

I have a similar library for Perl, taken from the Perl 4 man page.
(I finally got around to removing the &s from the procedure names.)
But the point is that it should work with the regular open() system
call, such that calling open("/dev/tcp/<host>/80", O_RDWR) should open
host "<host>" on port 80, and something like "serv(80, cookie)"
should copy a string into cookie such that open(cookie, O_RDWR) would
accept a connection.  Unfortunately, C makes it very hard to
override the meaning of global function names cleanly.

-- 
John Cowan          http://www.ccil.org/~cowan        cowan at ccil.org
Even a refrigerator can conform to the XML Infoset, as long as it has
a door sticker saying "No information items inside".  --Eve Maler

[TUHS] networking on unix before uucp

[Dave Horsfall]

On Fri, 22 Aug 2014, Larry McVoy wrote:

> > In a previous slavery we had a simple socket library; the application 
> > did little more than say whether it was a client or a server etc.  I 
> > wish I could steal that code now...
> 
> lmbench has one:

Yeah, that's pretty much what we did.  Conceptually, the client said 
"please connect me to Sydney's pogo stick server", and the server said "I 
am the Sydney pogo stick server".  Networking as it ought to be (none of 
this business about the right phase of the moon and looking for a few 
all-too-scarce virgins etc).  The config stuff was initially in flat 
files, and it was my job to convert the lot to OpenLDAP lookups.

-- Dave

[TUHS] networking on unix before uucp

[Larry McVoy]

On Sat, Aug 23, 2014 at 06:37:59AM +1000, Dave Horsfall wrote:
> On Fri, 22 Aug 2014, Larry McVoy wrote:
> 
> > > In a previous slavery we had a simple socket library; the application 
> > > did little more than say whether it was a client or a server etc.  I 
> > > wish I could steal that code now...
> > 
> > lmbench has one:
> 
> Yeah, that's pretty much what we did.  Conceptually, the client said 
> "please connect me to Sydney's pogo stick server", and the server said "I 
> am the Sydney pogo stick server".  

If anyone wants the stuff we use, the stuff mentioned above, I can put it
up on the web.
-- 
---
Larry McVoy            	     lm at mcvoy.com             http://www.mcvoy.com/lm 

[TUHS] networking on unix before uucp

[Larry McVoy]

On Sat, Aug 23, 2014 at 07:01:40AM +1000, Dave Horsfall wrote:
> On Fri, 22 Aug 2014, Larry McVoy wrote:
> 
> > If anyone wants the stuff we use, the stuff mentioned above, I can put 
> > it up on the web.
> 
> Pretty please!  For private use only, of course.

You'all can use it anywhere you like.

http://www.mcvoy.com/lm/tcp.shar

It's not that big a deal (other than 20 years of bug fixes :)

Somewhere I have a bigger deal, at least I think it is, I made a library
to talk to Sun RPC servers in parallel.  I called it rpc vectors and Ron
Minnich used it to put a bunch of nfs servers together, he called that
bigfoot.  Paper below, if someone wants that code I can ship that too.
It was pretty neat, back in the days of 10Mbit ethernet I was querying
thousands of machines in a single call.  The code dealt with the fact
that you had to start eating the replies before you were done sending
the question :)

http://wenku.baidu.com/view/797c4ac62cc58bd63186bd1c.html

or for old school people

http://www.mcvoy.com/lm/bitmover/lm/papers/bigfoot.ps

The code was pretty small, pretty clever, it's a shame it didn't catch on.
-- 
---
Larry McVoy            	     lm at mcvoy.com             http://www.mcvoy.com/lm 

[TUHS] networking on unix before uucp

[John Cowan]

Dave Horsfall scripsit:

> Well, we mustn't imitate Unix, must we?

I shall dub Windows "WIUB", for "WIUB Imitates Unix, Badly".  (Cf.
LIAR, the compiler for MIT Scheme, whose name means "LIAR Imitates
Apply Recursively".)

-- 
John Cowan          http://www.ccil.org/~cowan        cowan at ccil.org
Awk!" sed Grep. "A fscking python is perloining my Ruby; let me bash
    him with a Cshell!  Vi didn't I mount it on a troff?" --Francis Turner

[TUHS] portal daemon (was Re: networking on unix before uucp)

[Aaron J. Grier]

On Fri, Aug 22, 2014 at 11:57:01AM -0400, John Cowan wrote:
> #     filedes = open( "/dev/net/harv",2 );
> #     if( filedes < 0 )
> #        printf(" harvard is dead");
> #     else
> #        while( (nbytes=read(filedes,buf,80)) > 0 )
> #           write( 0,buf,nbytes );                                   
> 
> If only this code still worked on modern Unixes!  The socket API is
> fine, but there really was no need to break good old open, at least
> for client-side operations.  Plan 9 got it right here, as usual.

was this the impetus behind mount_portal in 4.4BSD?  It's still
available in NetBSD, although I've never played with it myself.

-- 
  Aaron J. Grier | "Not your ordinary poofy goof." | agrier at poofygoof.com

[TUHS] portal daemon (was Re: networking on unix before uucp)

[Clem Cole]

Mumble -- I would say that BSD portals was more in response to an MIT idea
that the Apollo guys (Jim Rees I believe) put in domain as the Apollo Typed
File System.   At the time of the BSD, there was a lots of discussion going
on about how to move things out the kernel and a number of techniques were
considered -- Portals was one of them.


The idea of /dev/mumble/some_other_mumble was kicking around early UNIX for
a long time.   I think I first encountered it in the MIT Chaos code, but
the Arpanet NCP code may have redated its use.   The idea was simple, the
kernel's nami() function that was called by kernel open code (or any other
call the had to walk a pathname), left the pointer to the next undecoded
part of the path in the user's input buffer alone.   So a hack was put in
the kernel that caused open to vector to needed helper code for the network
to do the rest of the pathname to get the parameters.    The advantage of
this of course is open returns a real file descriptor.

BTW: One idea I had many years ago, and I remember talking to Dennis about
it at a couple of USENIX was similar to Apollo's typed scheme.   As dmr put
it to me at the time, UNIX has a typed file system with very few types.
 The question was if we opened up the typing scheme to allow for arbitrary
types, would you get more of mess or would the new feature be of an
advantage.  I was not doing research at the time - I was employed at
Masscomp and just getting a distributed FS was enough.  Since I did not
have a good use for the hack for EFS, I never tried to implement it.  It
was always on my "TODO" list.   Maybe when I retire and I can hack just for
the fun of it, I can try it ;-)

Clem


On Tue, Sep 2, 2014 at 3:15 PM, Aaron J. Grier <agrier at poofygoof.com> wrote:

> On Fri, Aug 22, 2014 at 11:57:01AM -0400, John Cowan wrote:
> > #     filedes = open( "/dev/net/harv",2 );
> > #     if( filedes < 0 )
> > #        printf(" harvard is dead");
> > #     else
> > #        while( (nbytes=read(filedes,buf,80)) > 0 )
> > #           write( 0,buf,nbytes );
> >
> > If only this code still worked on modern Unixes!  The socket API is
> > fine, but there really was no need to break good old open, at least
> > for client-side operations.  Plan 9 got it right here, as usual.
>
> was this the impetus behind mount_portal in 4.4BSD?  It's still
> available in NetBSD, although I've never played with it myself.
>
> --
>   Aaron J. Grier | "Not your ordinary poofy goof." | agrier at poofygoof.com
> _______________________________________________
> TUHS mailing list
> TUHS at minnie.tuhs.org
> https://minnie.tuhs.org/mailman/listinfo/tuhs
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20140902/f96dccb7/attachment.html>

[TUHS] portal daemon (was Re: networking on unix before uucp)

[Erik E. Fair]

Plan 9 went that way in a big way, as I recall the published papers
given at USENIX; they took naming in the UNIX filesystem to a high art.

	oh, and by the way, hi all,

	Erik Fair

[TUHS] networking on unix before uucp

[Dave Horsfall]

Does ACSnet (Australian Computer Science Network) count?  It ran over 
leased lines, supporting remote login and file transfer, back around the 
70s.  Definitely V6, because of the 11/40s at the time.  A commercial 
offshoot was MHSnet, which for all I know is still available.

-- Dave (dave:csu40)