Fw: Has Oz Come Up in This Forum?

[Peter Van Roy]

Dear Ken and Mark,

Here is some more information about the Oz language and the Mozart system.

> ----- Original Message -----
> From: Mark S. Miller <markm@caplet.com>
> To: Paul Snively <psnively@webeasy.com>
> Cc: <e-lang@eros-os.org>; Ken Kahn <ken@toontalk.com>
> Sent: Friday, January 21, 2000 3:22 PM
> Subject: Re: Has Oz Come Up in This Forum?
> 
> 
> > At 01:13 PM 1/19/00 , Paul Snively wrote:
> > >Folks,
> > >
> > >Has anyone here looked at Oz? <http://www.mozart-oz.org>
> > >
 ...
> > >
> > >This sure sounds like Trusty Scheme or E to me.
> >
> > Oz does indeed look very interesting.  It seems to share these properties
> > with E by virtue of common ancestry.  Oz is at least closely related to Flat
> > Concurrent Prolog, and probably derived from it.  As shown on
> > http://www.erights.org/history/index.html , Concurrent Prolog, Actors, and
> > KeyKOS all have capability nature, and together begat Joule.  Most of the
> > important formal properties of E, especially capability security, are
> > inherited directly from Joule.

Yes, Concurrent Prolog, Flat Concurrent Prolog, and related languages are
ancestors of Oz.

 ...
> >
> > In any case, I skimmed their web site and couldn't find the introduction you
> > quote above, or anything else on capability security.  Although they do
> > distributed programming, I didn't see anything to indicate they are doing
> > cryptographically-based distributed capabilities.  If not, then their above
> > paragraph would not hold between mutually suspicious machines.
> >
> > >Anyone know more about the Mozart environment and/or the Oz language?
> >

This is hard to summarize in a few sentences, since Mozart is the result of
a decade of work by many dedicated researchers.  Here's a try.  The Mozart
platform's main strengths are open fault-tolerant distributed programming
and constraint-based inferencing.  The basic philosophy of the Oz design is
to provide maximum expressivity, always keeping an efficient implementation
and a simple formalization.

The current Mozart release supports open network-transparent distribution
(language semantics independent of distribution structure) and fault 
tolerance.
While the language supports capability-based security, the current network
layer doesn't yet enforce it with cryptographic protocols.  The extensions
needed to the network layer are designed and will be put in a forthcoming
release.  Seif Haridi can give more information about this design.

Because of the factorized design of Oz, we provide network-transparent
distribution while allowing the programmer to keep full control over
network communication patterns.  To be precise, the Oz execution model
distinguishes between stateless, stateful, and single assignment data.  Each
of these is implemented by a different class of distributed algorithms.
This seems to work well in practice.

For more information, please look at the Mozart publications page
(http://www.mozart-oz.org/papers/).  Here are a few selections.

- Introduction to efficient network-transparent distribution, with examples:
  http://www.mozart-oz.org/papers/abstracts/diplcl99.html (short article)
  http://www.info.ucl.ac.be/people/PVR/talk240100.ps (slides of talk)

- In-depth overview of distribution, fault tolerance, and security:
  http://www.mozart-oz.org/papers/abstracts/ngc98.html

- Abstraction building in Oz (from a semi-formal viewpoint):
  http://www.mozart-oz.org/papers/abstracts/volume1000.html

- Oz for logic programmers and Prolog people: (also "lessons learned")
  http://www.info.ucl.ac.be/people/PVR/tut210100.ps

- Design philosophy of "language-based fault tolerance":
  http://www.mozart-oz.org/papers/abstracts/sendai99.html

Best regards,

Peter

-- 
Prof. Peter Van Roy
Dept. of Computing Science and Engineering
Université catholique de Louvain
B-1348 Louvain-la-Neuve, Belgium

Email: pvr@info.ucl.ac.be
Tel: (+32) (10) 47.83.74
Web: http://www.info.ucl.ac.be/~pvr/
Mozart: http://www.mozart-oz.org