Fw: Has Oz Come Up in This Forum?

[Mark S. Miller]

At 02:07 AM 1/24/00 , Seif Haridi wrote:
>This email is to Mark.

Hi Seif, if you want to contact me privately, I can be reached at 
markm@caplet.com.  My preference is that correspondence of general E 
interest include the list, so I am doing so in my response.

>Mark, you know me! I used to cooperate with Ken and Vijay Saraswat, and I wanted
>to contact you for while! I am codesigner of Oz, and co-designer of distributed
>computing
>support (together with Per Van Roy and Per Brand).

Yes, I remember you too!  Good to hear from you again.

>First Oz is a direct descendent of concurrent logic/constraint programming where
>lots of shortcommings there were removed.
>
>We are currently working on SECURE GLOBAL computing is Mozart (where encryption will
>be used). Ken has suggested for a while to contact you, and it seems this is the
>right time.
>I will send you soon my slides on our recent project started this year called SEFTS
>(SEcurity, Fault-Tolerance and Scalibility in distributed computing) where you will
>see
>our current thinking.
>Shortly the current state of Mozart is that it supports a lot of security issues and
>resource
>control IF things are written in Mozart, and some things for malicious users (e.g.
>code verification). The next step is to handle hostile environments and that is what
>SEFTS is
>about.
>
>I would like to get in touch with you to discuss further!

It sounds like it will be fascinating to compare approaches.  We look 
forward to hearing more.  Given your starting point, I suspect we have 
similar views of concurrency & distribution.  However, I'm quite curious to 
hear your views on security & persistence.  Our view of security is 
summarized well by http://www.erights.org/elib/capability/ode/ .  Our view 
of persistence is not yet stated clearly anywhere (perhaps our discussion 
will prod me into doing so), but I leave you for now with the mysterious 
(and over-simplified) comment that I no longer believe in orthogonal 
persistence.

Another general issue who's importance is only now becoming clear to me is 
distributed linking under mutual suspicion.  I explain the problem in 
http://eros.cis.upenn.edu/%7Emajordomo/e-lang/1264.html and 
http://eros.cis.upenn.edu/%7Emajordomo/e-lang/1270.html .  The 
not-yet-written Part 3 will explain the solution I have in mind.  Since the 
problem seems universal, I'm wondering if you have an approach to this as well.

A potentially useful comparison exercise to explore: Since we both have 
small kernel languages, we might want to see what it looks like to implement 
either kernel in the other.  This is usually a good way to start comparing 
expressiveness issues.  I currently don't see the "constraint-based 
inferencing" as relevant to E's goals (though I'm certainly willing to let 
you try to persuade me).  OTOH, the "fault tolerance", if achieved in a way 
consistent with our other goals, would be *extremely* valuable to us.


         Cheers,
         --MarkM