[TUHS] 32I status as of 17 Nov 2003

Brantley Coile bwc at coraid.com
Tue Nov 18 07:41:30 AEST 2003


On Mon, 17 Nov 2003 11:40:24 EST, <patv at monmouth.com> wrote:

> 1. Converted unpublished locore to AT&T syntax.  Work still underway on
> new locore.S file.
>
> 2. Will use coff executable format. May comply with Sys V ABI.  The
> latter is TBD.

Please don't do that.  There is no value in being Sys V ABI compliant and
the original a.out work very well.  I've used it for two and a half
decades now, and still use it on Plan 9.  There just needs to be magic
numbers for the various formats.  I would suggest looking at the
10Edition information for things that arn't as they currently are.

>
> 3. Gcc is tool chain unless someone wants to take on 32V compiler.  May
> take advantage of inline and asm in machine dependent files.
>
> I wouldn't recommend using the 32V compiler, mainly because there
> are so many syntax changes, as well as newer compiler technology,
> missing from it as to make it a very challenging project.

There are a couple of options here.  One is to use one of the compilers
from MIT that will match the code in 32V very closely.  You can google for
these if they aren't already on Warren's site.  The other option
is to use LCC.  I wouldn't use GCC if I were doing it.  (Of course I'm not
so these are just suggestions.)
>
> 4. Gcc cross compile to be used for building.  I'll make this available
> later next month, after I shake out the bugs.
>
> 5. May go to Bitkeeper for source control. More to follow.
>
> 6. License for all new files to be the "revised BSD license."  This is
> compatible with the Caldera license, an "original BSD license" with
> the advertising clause specific to Caldera.  I want to keep it open
> source, and this is the best compromise I see for project license.
>
> 7. May change spln() to more understandable nomenclature, e.g., spl4()
> becomes spltty(), similar to BSD practice.
>
> 8. 32I is the interim name.  I would have preferred Unix version 7, but
> can't for obvious trademark reasons.

I'm not sure if the current trademark owners can retro actively
disassociate a mark after the fact.  If it WAS Unix 32V it still IS Unix 
32v.
I suppose if you do much more than port it to Intel one could argue
that it's not Unix 32V anymore.

>
> Project name still up for grabs.  I was thinking of UNX, named after
> the old DEC name for the facility I work in.  Sort of a tribute to
> days gone by, as is porting 32V.  Probably get into trouble for that
> one as well.
>

  Brantley



More information about the TUHS mailing list