[TUHS] So now that the source is finally out...

[TUHS] So now that the source is finally out...

[John Holden]

> > I'm fairly sure things like "=+" and so on were replaced with "+="
> > in the move from V6 to V7.
> 
> I thought so too, but I checked, and there are some still in there.

I think you'll find that the kernel code is clean, but a lot of the older
utilities will show their earlier lineage (including the C compiler)

[TUHS] So now that the source is finally out...

[Ken Wellsch]

On Sun, Jan 27, 2002 at 02:03:40PM -0500, norman at nose.cs.utoronto.ca wrote:
> 
> > Ken Wellsch:
> > I'm fairly sure things like "=+" and so on were replaced with "+="
> > in the move from V6 to V7.
> 
> The V7 C compiler accepted =+, but it still accepted += as well; there was
> a lot of code written the old way, and nobody wanted to be forced to convert
> everything all at once.  

Apologies, typical of my terse replies - I was actually concerned
with clearing up the possible misconception that "=+" was a "feature"
of V7 rather than an "obsolete" holdover from V6.  I suppose historic
accuracy in this context is of little use, I don't know.  I certainly
can believe one can find such artifacts in the existing V7 code 8-)

Cheers,

-- Ken

[TUHS] So now that the source is finally out...

[norman]

Ken Wellsch:

  I'm fairly sure things like "=+" and so on were replaced with "+="
  in the move from V6 to V7.  I think structure assignments were added
  here too, and the much more obscure, being able to declare passed
  arguments in the function preamble as "register."  I believe K&R
  reflects the C language as seen with V7...

The V7 C compiler accepted =+, but it still accepted += as well; there was
a lot of code written the old way, and nobody wanted to be forced to convert
everything all at once.  Similarly, the V7 compiler complained about implicit
conversions between pointers of different types, or between ints and pointers,
but they were treated as warnings, not fatal errors; you could still compile
old code and just ignore the compiler's fussing.

I recall that when I arrived at Bell Labs in 1984, it was apparently not long
after the research group's C compiler had been changed to treat all the
obsolete stuff as errors; certainly there was still code in /usr/src that
hadn't been updated, and all of it had been recompiled recently to run on
the VAX.

A modern C compiler would choke even on some of the stuff that was legal
in those days.  Recently I recompiled tbl with lcc, and had a grand time
cleaning out all the ideas we all thought were clever in the late 1970s
but most of us would never think of doing now.  For those with the virgin
source handy, take a look at subroutine `point' and the way it is used
and abused.

Norman Wilson

[TUHS] So now that the source is finally out...

[Ken Wellsch]

> On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
>
> gcc can compile K&R, but the language has evolved some since the v7
> days.  =*, =+, etc became *=, +=, etc.  There are some other subtle
> things too that I don't recall off the top of my head, but which vexed
> the comp.lang.c news groups in the early 1980's.

I'm fairly sure things like "=+" and so on were replaced with "+="
in the move from V6 to V7.  I think structure assignments were added
here too, and the much more obscure, being able to declare passed
arguments in the function preamble as "register."  I believe K&R
reflects the C language as seen with V7...

-- Ken

[TUHS] So now that the source is finally out...

[Greg Lehey]

On Saturday, 26 January 2002 at  9:50:21 -0500, Ken Wellsch wrote:
>> On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
>>
>> gcc can compile K&R, but the language has evolved some since the v7
>> days.  =*, =+, etc became *=, +=, etc.  There are some other subtle
>> things too that I don't recall off the top of my head, but which vexed
>> the comp.lang.c news groups in the early 1980's.
>
> I'm fairly sure things like "=+" and so on were replaced with "+="
> in the move from V6 to V7.

I thought so too, but I checked, and there are some still in there.

Greg
--
Finger grog at lemis.com for PGP public key
See complete headers for address and phone numbers

[TUHS] So now that the source is finally out...

[Joerg Micheel]

On Mon, Jan 28, 2002 at 10:28:29AM +1030, Greg Lehey wrote:
> > I'm fairly sure things like "=+" and so on were replaced with "+="
> > in the move from V6 to V7.
> 
> I thought so too, but I checked, and there are some still in there.

Indeed, I remember those well, when we had to bootstrap the compiler
on a PDP-11 running RSX-11M.

	Joerg

[TUHS] So now that the source is finally out...

[Carl Lowenstein]

> From: "P.A.Osborne" <P.A.Osborne at ukc.ac.uk>
> To: "M. Warner Losh" <imp at village.org>
> Cc: tuhs at minnie.tuhs.org
> Subject: Re: [TUHS] So now that the source is finally out...
> Date: Fri, 25 Jan 2002 11:16:37 +0000
> 
> On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
> > Real as in 286 or as in 8088 :-).  
> 
> 
> > gcc can compile K&R, but the language has evolved some since the v7
> > days.  =*, =+, etc became *=, +=, etc.  There are some other subtle
> > things too that I don't recall off the top of my head, but which vexed
> > the comp.lang.c news groups in the early 1980's.
> 
> That makes things a challenge.  Still the source of the kernel is
> around 10K lines IIRC and going through it in stages doesnt make life
> too painfull.

Doesn't one of the C beautifiers do that for you (rewrite =+ etc.)
Either GNU indent or Berkeley? cb.

    carl
-- 
        carl lowenstein         marine physical lab     u.c. san diego
                                                  clowenstein at ucsd.edu

[TUHS] So now that the source is finally out...

[Rob Becker]

who is going to port 7th edition to the i386? (:
Rob Becker

[TUHS] So now that the source is finally out...

[P.A.Osborne]

On Thu, Jan 24, 2002 at 01:39:50PM -0500, Rob Becker wrote:
> who is going to port 7th edition to the i386? (:

OK I know that the comment was a joke but...

Funnily enough I started looking at doing just that around 18 months 
ago.  Unfortunately (from that point of view) it ground to a halt when
wedding arrangements took priority.   

Anyhow the wedding is out of the way - the missus is now a computing 
student so I should get some sympathy from her (she likes Unix :-), so
hopefully I can resurrect my notes etc and start working on it again. 

No promises mind as I havent touched C or assembly for around 18months
now so am getting increasily rusty.  Not that my x86 assembler is 
particulary good,  or that the missus will allow me enough free time to
get my head back into coding at home...

So the following is what I can remember from quite a while ago...

Also from what I did look at,  I rapidly came to the conclusion that 
a "real" mode i386 port was the more straight forward as that keeps
everything in 16 bits and saves a lot of mucking around.  

So you could either rewrite it as a process that runs as 16 bits and
so use it as an app under say Linux/FreeBSD/OS/2  (I wont mention the
people from Redmond :-),   or as a 16 bit OS running native.

Either way the first skirmishes I made with the source lead me to 
believe that in simple terms, the following would need to be completed:
-  rewrite the assembler portion to x86 - memory management/interrupt
   handlers/booting and initialisation etc etc
-  rewrite the disk driver to run the floppy drive (1.4 MB should be
   enough to demonstrate it works and boot a kernel etc)
-  the tty stuff could conceivably be pointed at COM1 and you could
   hang a console/teletype (for the sad) onto the back of a PC  or
   just produce suitable keyboard/screen drivers.

With any joy if the assembler level can be done correctly (device drivers
aside) you could theoretically leave the bulk of the C source intact - which
could possibly make it a purer port, assuming you can get gcc to compile 
K&R C - I havent tried yet.


Anyhow comments and sarcastic remarks are welcome.

Regards

Paul


------------
Paul Osborne
Computing Officer
University of Kent at Canterbury Computing Service

[TUHS] So now that the source is finally out...

[M. Warner Losh]

In message: <20020125094218.C5968 at apple.ukc.ac.uk>
            "P.A.Osborne" <P.A.Osborne at ukc.ac.uk> writes:
: Also from what I did look at,  I rapidly came to the conclusion that 
: a "real" mode i386 port was the more straight forward as that keeps
: everything in 16 bits and saves a lot of mucking around.  

Real as in 286 or as in 8088 :-).  I've love to have a v7 port to my
DEC Rainbow 100, which isn't too IBM-PC-like other than the fact that
both have an 8088.  There was a 8088 port of V7 (or maybe it was v6)
to the Rainbow marketed under the name of venix, but no one seems to
be able to find a distribution anymore.

: With any joy if the assembler level can be done correctly (device drivers
: aside) you could theoretically leave the bulk of the C source intact - which
: could possibly make it a purer port, assuming you can get gcc to compile 
: K&R C - I havent tried yet.

gcc can compile K&R, but the language has evolved some since the v7
days.  =*, =+, etc became *=, +=, etc.  There are some other subtle
things too that I don't recall off the top of my head, but which vexed
the comp.lang.c news groups in the early 1980's.

Warner

[TUHS] So now that the source is finally out...

[P.A.Osborne]

On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
> Real as in 286 or as in 8088 :-).  

Yup,  otherwise you start mucking around with the protected mode
shuffle to 32 bits and I have been burnt by that before.  The ix86
chips are startup from cold in 16bit mode so it makes life easier.

> I've love to have a v7 port to my
> DEC Rainbow 100, which isn't too IBM-PC-like other than the fact that
> both have an 8088.  There was a 8088 port of V7 (or maybe it was v6)
> to the Rainbow marketed under the name of venix, but no one seems to
> be able to find a distribution anymore.

Oooh that would be so usefull and save so much mucking around.

> gcc can compile K&R, but the language has evolved some since the v7
> days.  =*, =+, etc became *=, +=, etc.  There are some other subtle
> things too that I don't recall off the top of my head, but which vexed
> the comp.lang.c news groups in the early 1980's.

That makes things a challenge.  Still the source of the kernel is
around 10K lines IIRC and going through it in stages doesnt make life
too painfull.

I must be mad thinking about this again....


paul

[TUHS] So now that the source is finally out...

[Ephrem Hugh Bensusan]

It seems that there is a distribution of Venix hanging about, but I
believe it is for the DEC PRO 350 or 380.  So I don't know how useful 
it might be in this context.

In any case, one can find it at:
http://www.os2site.com/sw/dec/pro/venix/

-- 
Ephrem Hugh Bensusan                    EMail: ephrem at bensusan.net
                                        Phone: 352.687.3757


-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GIT/CS d? s--:+ a C++++$ UB/L++++$ P++> L+++$ E W++(+++)$ N++> o? K++ w-- 
O M? V-- PS+++(--) PE Y+ PGP+ t++* 5- X++ R tv- b+++> DI-- D+ 
G++ e* h----(-)* r+++ y++++(+++(--)) 
------END GEEK CODE BLOCK------

On Fri, 25 Jan 2002, P.A.Osborne wrote:

> On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
> > Real as in 286 or as in 8088 :-).  
> 
> Yup,  otherwise you start mucking around with the protected mode
> shuffle to 32 bits and I have been burnt by that before.  The ix86
> chips are startup from cold in 16bit mode so it makes life easier.
> 
> > I've love to have a v7 port to my
> > DEC Rainbow 100, which isn't too IBM-PC-like other than the fact that
> > both have an 8088.  There was a 8088 port of V7 (or maybe it was v6)
> > to the Rainbow marketed under the name of venix, but no one seems to
> > be able to find a distribution anymore.
> 
> Oooh that would be so usefull and save so much mucking around.
> 
> > gcc can compile K&R, but the language has evolved some since the v7
> > days.  =*, =+, etc became *=, +=, etc.  There are some other subtle
> > things too that I don't recall off the top of my head, but which vexed
> > the comp.lang.c news groups in the early 1980's.
> 
> That makes things a challenge.  Still the source of the kernel is
> around 10K lines IIRC and going through it in stages doesnt make life
> too painfull.
> 
> I must be mad thinking about this again....
> 
> 
> paul
> _______________________________________________
> TUHS mailing list
> TUHS at minnie.tuhs.org
> http://minnie.tuhs.org/mailman/listinfo/tuhs
> 

[TUHS] So now that the source is finally out...

[M. Warner Losh]

In message: <Pine.GSO.4.21.0201251013350.610-100000 at barclay>
            Ephrem Hugh Bensusan <ephrem at bensusan.net> writes:
: It seems that there is a distribution of Venix hanging about, but I
: believe it is for the DEC PRO 350 or 380.  So I don't know how useful 
: it might be in this context.
: 
: In any case, one can find it at:
: http://www.os2site.com/sw/dec/pro/venix/

I think this is binary only...  Maybe it would make a good addition to
the TUHS archive, however...

Warner

[TUHS] So now that the source is finally out...

[David C. Jenner]

This Venix has been available from Barry Kort and ftp.update.uu.se/professional
for many years.  It is "Pro/Venix" versions 1 and 2.  Pro/Venix came directly
from VentureCom, the vendor who ported it.  More desirable is "Venix/Pro"
especially version 2.  Venix/Pro came from DEC, who got it from VentureCom and
spiffed it up a bit for the Pro.  Anyone who has or comes across floppies for
Venix/Pro from DEC should really archive them (and please let me know!  I have
a set of floppies with one or two bad disks).

Whether this is "legal" or not is up to some question.  VentureCom put Pro/Venix
out in the public domain, but the only "proof" of this is its widely available
existence and a verbal statement from the person who first got it.  At some
point in the past, I think Bob Supnik thought it was reasonable that DEC would
put its interests in the product in the Ancient Unix license, but he no longer
works for DEC.  There is also some question whether this Venix was ported from
System III, which would eliminate it from the Ancient Unix license and the new
form of it.

Dave

"M. Warner Losh" wrote:
> 
> In message: <Pine.GSO.4.21.0201251013350.610-100000 at barclay>
>             Ephrem Hugh Bensusan <ephrem at bensusan.net> writes:
> : It seems that there is a distribution of Venix hanging about, but I
> : believe it is for the DEC PRO 350 or 380.  So I don't know how useful
> : it might be in this context.
> :
> : In any case, one can find it at:
> : http://www.os2site.com/sw/dec/pro/venix/
> 
> I think this is binary only...  Maybe it would make a good addition to
> the TUHS archive, however...
> 
> Warner
> 
> _______________________________________________
> TUHS mailing list
> TUHS at minnie.tuhs.org
> http://minnie.tuhs.org/mailman/listinfo/tuhs

-- 
David C. Jenner
djenner at earthlink.net

[TUHS] So now that the source is finally out...

[Wolfgang Rupp]

On Friday 25 January 2002 12:16, P.A.Osborne wrote:
> On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
> > There was a 8088 port of V7 (or maybe it
> > was v6) to the Rainbow marketed under the name of venix, but no one
> > seems to be able to find a distribution anymore.
>
> Oooh that would be so usefull and save so much mucking around.

ftp://ftp.update.uu.se/pub/professional/venix

HTH

Wolfgang Rupp

[TUHS] So now that the source is finally out...

[M. Warner Losh]

In message: <200201251624.g0PGOd101040 at tk212017121038.teleweb.at>
            Wolfgang Rupp <rupp at chello.at> writes:
: On Friday 25 January 2002 12:16, P.A.Osborne wrote:
: > On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
: > > There was a 8088 port of V7 (or maybe it
: > > was v6) to the Rainbow marketed under the name of venix, but no one
: > > seems to be able to find a distribution anymore.
: >
: > Oooh that would be so usefull and save so much mucking around.
: 
: ftp://ftp.update.uu.se/pub/professional/venix

This is not the Rainbow version, but instead the PRO version.  I'd
love to see the DEC Rainbow (or even the IBM PC) verions available.

Warner

[TUHS] So now that the source is finally out...

[Wolfgang Rupp]

On Friday 25 January 2002 12:16, P.A.Osborne wrote:
> On Fri, Jan 25, 2002 at 03:27:25AM -0700, M. Warner Losh wrote:
> > I've love to have a v7 port to my
> > DEC Rainbow 100

oops... I only read "Venix" and jumped to conclusions.
Sorry. 

However, under /pub/rainbow there is rainbow stuff
on ftp.update.uu.se too.


Wolfgang Rupp

[TUHS] So now that the source is finally out...

[Jeffrey S. Sharp]

On Thursday, 24 Jan 2002, Rob Becker wrote:

> who is going to port 7th edition to the i386? (:

Well, I surely wouldn't mind if people made it work on more PDP-11s than
just a /45 or /70 with RP disks.

--
Jeffrey S. Sharp
jss at subatomix.com