On 12-Jan-06, at 4:32 PM, Bradley Lucier wrote:
Bill Clementson talks about concurrent "functional" programming in his blog, and laments the lack of public progress on Termite:
Since the Termite paper was presented at the European Scheme and Lisp workshop a few things have happened:
1) The mailbox mechanism has been integrated into Gambit's thread model, which improves the performance of messaging in Termite, and allows "message-passing" programming (between local threads) in the standard Gambit system. Serialization and deserialization has also been improved.
2) A "Distributed Computing" example was added to the examples in the distribution. It shows how to implement a distributed computing library in about 700 lines of Scheme code (including process migration, location transparency, and serialization of I/O ports).
3) Guillaume Germain will soon be done with his thesis, and has written several interesting Termite examples. Stay tuned.
The most difficult aspect of this distributed programming language research is that there are so many abstraction levels to choose from. Should Termite be a minimal language on which others can build their own distributed computing system, or should it provide a large set of predefined features (such as process migration, location transparency, etc)? In the spirit of Scheme, Termite aims to be a small maleable distributed programming language. We're hoping that it will be the basis for other more specific distributed programming languages developed here or elsewhere.
Marc