below
On Fri, Apr 24, 2015 at 1:04 PM, Dave McGuire <mcguire at neurotica.com> wrote:
I don't know if calling it "insane" is reasonable, Sampsa. You've
fired up a thirty-year-old OS and are expecting it to act like it's a
current OS. It won't.
?Amen?
I ran Ultrix-32 daily back when it was current. It was (and is) a
great OS.
?+1? I was the Technical Lead for Ultrix 4.4. If you can get anything
post 4.4 you are likely to be much happier. Ultrix 4.3 <<< Ultrix 4.4 --
a huge number of fixes and enhancements went into 4.4 - which was where the
MIPS 4000/4400 was first supported. I compiler was upgraded and a whole
SCSI I/O system was developed for it that was common with what would
eventually be Tru64 (there was some discussions of merging the projects -
but that never happened - post Compaq).
So you might find it a little "fresher" in your mind.
But as with all commercial UNIX implementations,
everyone
fixes a bunch of stuff as their first task after installation. For
Ultrix-32, nearly everyone installs GNU Tar, Bash, and GCC immediately.
?Hmmm - ?yes and no. I personally never install bash on anything -- i.e.
as a BSD developer (I believe in typing to the csh and programming the
Bourne Shell). But I see your point. There are tools that people move
because they are comfortable with them. I also don't see the point in GCC
since the DEC VAX compiler was pretty good in those days the gcc did not
generate as good code. But if you wanted/needed something that gcc
supported and was not yet in DEC C, it might be worth it.
Its interesting Sampsa never programmed with a "white book" C compiler
--hey - as an old f*rt - I learned C the real way -- talking to dennis and
eventually reading his paper ;-) [I used to have a xerox of the proofs for
the whitebook - I remember when Dennis and Brian were writing it]. That
said, I suspect if you grew up with "modern" C, the original would feel a
tad naked.
Note: I once pointed out to Dennis that that boot loader for System V had
grown to be larger that the V6 kernel. This was around the time when
Rob Pike wrote the "cat -v considered harmful" paper. Holzmann of MIT &
JPL takes a stab at this in a new very short paper:
http://www.computer.org/cms/Computer.org/ComputingNow/issues/2015/04/mso201…
where he discusses "software inflation."
The same goes for SunOS4 and HP-UX, from the same era.
?As I said, this is because people are comfortable with some specific tools
and want them. The problem is if you never learned to us ed (or EDT on a
VAX/VMS) then it is a b*tch to land on a system that offers you something
you don't know how to use. So you immediately move or write a version of
an old tool. Paul Cantrel's teco (which is pretty much the universal C
implementation of same that is flying around these days) came to be because
he did not like vi or emacs and had come to UNIX from the 10s and the VMS.
So he wrote his own (we also wrote a set of emacs macros to emulate EDT for
some of the other ex-VMS folks @ the time).
Actually I like that approach quite a bit better than, say, today's
Solaris for example.
?I agree. I have a image of a number of programs with a (bourne) shell
script that patches them so I can recompile a couple of tools I bring with
me on any system.?
? Sometimes to same tools are they but have been screwed with and command
line flags changed etc. The roms in fingers were programmed long ago. ;-)
Clem?