[virtmach] ETHZ Oberon/SlimBinaries [was Re: Is this thing...]

Dwight Hughes dwighth@ipa.net
Mon, 26 Jun 2000 21:25:16 -0500


I would place the ETHZ Oberon/SlimBinaries approach just outside the
realm of virtual machines. IIRC, you basically have a memory resident
compiler backend that performs on-demand compilation of the compact AST
intermediate representation to native code on a module by module basis
(it always loads and compiles a whole module at a time, regardless of
how much or little of the module is actually executed - though I think
it may start compiling the module as it is being loaded). With its
GC/memory management and OS abstractions it is somewhat more than the
usual compiled code plus support libraries, but not quite a VM, IMO. As
an environment though, the difference is rather difficult to spot in
use.

-- Dwight


btanksley@hifn.com wrote:
> 
> From: Denis N. Antonioli [mailto:antonio@ifi.unizh.ch]
> >On Mon, 26 Jun 2000 btanksley@hifn.com wrote:
> 
> >> Both Forth and Oberon are systems which are written to be
> >> compiled almost
> >> entirely to their own virtual machines.
> 
> >I have to disagree here.
> 
> >I've studied CS at the ETHZ, at the time they introduced Oberon for the
> >students' exercises and have taken a few courses given by Wirth on the
> >design of the system: Oberon was written to compile directly to native
> >code.
> 
> An interesting distinction.  No doubt Oberon was _designed_ to compile to
> native code.  But the fact remains that several Oberon systems are written
> for its own virtual machine, and compile to it.
> 
> >Or are you thinking of Franz's OMI (or SlimBianries)? But he stores
> >more or less the AST produced by the compiler. I don't think that
> >pass as a virtual machine, does it?
> 
> Now this is a very good question; I would agree that SlimBinaries are not a
> virtual machine, but I postulate that they _require_ a virtual machine in
> order to run.  If there were no VM in Oberon, then SlimBinaries would not be
> portable.
> 
> The problem is that the VM isn't well-specified, not that it doesn't exist.
> 
> >       dna
> 
> -Billy