At 12:34 AM 3/15/2003 -0500, Robertdkeys@aol.com wrote:
>As a starting point, I would like to see a compilable and usable
>pcc port.  Since all of the antique unix is now BSD licensed,
>it should be compatible.... and usable.

This is not true Robert, pcc is still AT&T proprietary and as the recent 
SCO lawsuits reveal (the current holder of the IP) there is someone out 
there still willing to sue people who use AT&T UNIX code.

Another negative is that many of the packages also use "gnuisms"

>As I see it, the issues are:
>
>1.  What level of NetBSD to use as a Classic VAX starting point
>      (I tend to think anything before elf and anything before rc.d).

But that would not be a good issue, ELF and particular shared libraries, 
give you a tremendous opportunity to "optimize" the working set that most 
systems use. Further the recent changes to the way memory segments were 
layed out allows a single process to use "all" of memory if it needs to. 
Those are good things that you'll want in your lean and mean OS.

>2.  What compilers are potentially usable, and available to work with
>      (I tend to pcc, or an earlier gcc maybe as early as 1.42 or so, if
>       it is workable on the VAX).

An earlier GCC would be your best bet, easiest to work with and likely free 
of IP entanglements.

>3.  To what hardware do we limit the Classic VAX (anything qbus and
>       earlier would be my expectation).

Is a 4000/705a Qbus? :-) Actually I think you just limit it to "VAX" and 
you will have _really fast_ VAXen and fun to use VAXen.


>4.  Is there any real interest in trying something like this?
>      (Who knows.... I just am getting this slowing feeling rather
>       that that"lovin' feelin'" on the VAX [as the old Righteous Bros.
>       song used to go...]).

There is probably interest, but it is hard for people to visualize where 
they might help. That is why these projects are most often successful when 
one individual starts cranking along and keeps the list of things that 
people could do wide open and well publicized. Unfortunately that "core" 
individual has to be a fairly good programmer and have a good feel for 
system issues.

>Am I nuts?  Yeah, mebbie, but.... ancient VAX is S l o w i n g,
>S  l  o  w  i  n  g..... down.....(:+}}... why should it need to.  It
>should not be heresy to say, "pick the best we can for it, and
>leave it at that.....

Well its a good kind of nuts. Another approach is to just take the current 
1.6x and start doing performance analysis on it and fixing those things 
that are slow. Understanding swapping behavior, code call chain lengths, 
what's emulated and what isn't. All those things can get you to a lot 
faster machine.

--Chuck