Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!vax1.physics.oxford.ac.uk!HALLAM
From: HALLAM@vax1.physics.oxford.ac.uk
Newsgroups: comp.sys.transputer
Subject: (none)
Message-ID: <8904111813.AA10346@uk.ac.ox.prg>
Date: 13 Apr 89 01:39:20 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 31
X-Unparsable-Date: 11-APR-1989 19:10:19 GMT +01:00

re Occam compilers - what language ?

   I am told that the origional Inmos Occam compiler was written in Occam, ie
Occam1 ! This was then extended to make Occam2 and then converted automaticaly
to Occam2. The result was not pretty.

  Inmos are now reputedly writting the compiler in C. This means they can port
it to any machine they feel like.

  I am also writting an Occam compiler, with additions to allow the simulation
of a large array of transputers (100 at first then 500). I would have liked to
write the program in Occam (we even have a Vax Occam compiler). However this 
program must interface with another (larger) simulation written in Fortran
(probably FortranIV !) hence a 'proper' VAX language is needed. The reason
I would like to use Occam is that the use of CHANnels allows you to build
all three phases of the compiler together in an object-y oriented-ish kind of a
manner. Howevever since a compiler is basicaly twiddling about with compilicated
recursive data structures I can't use Occam and I used C instead.

  O.K. I have read that Inmos note on recursive data structures - but so what.
I don't want to simulate data structures - too hard - too messy. That's the sort
of thing I want a compiler to do for me. I believe that until the Occam model
is extended to cover them Occam will remain restricted in it's application.

Finally a plug for the ZEUS experiment trigger processor - 500 processors
performing online data analysis on data from a particle accelerator 
detector chamber.

         Phillip Hallam-Baker

             Oxford University Nuclear Physics Dept