Oz and Mozart Hackers Mailing List

Re: 3 concrete goals for the short term


From: Denys Duchier (duchier@ps.uni-sb.de)
Date: Thu Mar 27 2003 - 09:16:58 CET


Here are some projects in no particular order (except the 1st one):

* PORT THE MODS FROM RELEASE BRANCH TO DEVEL TRUNK

  This is one for Kevin and me. Others are of course more than
  wellcome :-)

* INCREMENTAL TELL

  the lack of it means that when unifying datastructures containing
  futures we will retain the entire datastructures until all futures
  have been bound (with streams, this means for ever).

  I prototyped a version of incremental tell before, so I suppose I
  could take care of this.

* 64 BIT CLEANUP

  with Intel delaying their plans for 64 bit, this item is less urgent
  than it once seemed. However, there are 64 bit architectures and
  people have tried (of course unsuccessfully) to build Mozart on
  them. Another advantage would be to make the system more robust by
  removing implicit assumptions.

* DLL ISSUES, SYMBOL EXPORTS, ETC

  we need to resolve the issues with DLLs on Windows and we need to
  make it more practical to build extension libraries for Mozart. In
  particular this means exporting more symbols and more APIs (for
  example, there is no way to reuse Oz propagators in 3rd party native
  functors)

* REORGANIZATION

  the configure/build system has become too complicated to manage
  sanely. Leif and I have discussed it and we would like to see a
  reorganization into a _staged_ build, starting with a minimal core
  system. Libraries and tools should then be built using ozmake or
  something like it from a partial install. contrib must go (real
  contribs need to be maintained and made available separately).

* SECURITY

  We need to make it possible to execute untrusted applications and to
  compile untrusted code (there is currently no security with
  "prepare" sections in functors). Mozart intended to address
  security issues through module managers, but did not follow
  through. We should do this now. I do not wish to entertain
  alternative approaches to security; that's interesting too, but we
  should first try to make module managers practical for their
  intended purpose, and learn from the experience.

* SYSTEM THREADS

  We need an interface for system threads so that we can have
  e.g. native services servicing potentially blocking requests.
  I did a prototype for this too.

* CONSTRAINTS

  I am glad that Christian is back in the game. Indeed we need to
  improve things; e.g., while this is not my area of expertise, I
  understand that scheduling support is showing its age. Personally,
  I'd like to get back into improving/developing set/selection/tree/graph
  constraints. I'd like to see propagator priorities or something
  like that to help achieve reasonable staged propagation.

* REVAMP THE WEB SITE

  I think this is important because it is the face that we present to
  the world. I'd like a visitor to be able to find out quickly what's
  currently going on with Mozart (not just with us, but also in the
  user community). I think that the web site needs to _feel_ dynamic
  because that will also be the perception for Mozart itself.
  Christian mentioned the publication database; I'd like to accomodate
  publications not by us. I have mentioned teaching material before
  too. etc...

* DOCUMENTATION

  ozdoc is showing its age. Nowadays we should adopt the XML/XSLT
  combo. I have some preliminary bindings for libxml2/libxslt which
  we can use as the foundation of a more modern ozdoc.

Cheers,

-- 
Dr. Denys Duchier
Équipe Calligramme
LORIA, Nancy, FRANCE
-
Please send submissions to hackers@mozart-oz.org
and administriva mail to hackers-request@mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.



This archive was generated by hypermail 2b29.