Go backward to Motivation
Go up to Reorganizing Work Packages 4 and 5 into a single WP called "4/5"
Go forward to Task 4/5.2 : Distributed code generation
Task 4/5.1 : Fundamentals
As a result of the pressure from another Esprit project for getting
distributed code generation working properly (at least in restricted
situations for target architecture), a fundamental study started
autumn 1997, with first set of "unexpected" results presented at
the first review (This is recalled for the sake of completeness in
the list of deliverables below).
Further progresses occurred during 1998, and will be provided as a
deliverable. In fact, we expect to continue this set of
investigations, as interesting relations seem to emerge with abstract
interpretation.
Briefly speaking, the type of result we are interested in is the
following. Given a specification based on a synchronous paradigm,
- how and when can we preserve some kind of semantics when
relaxing the synchrony hypothesis, both for a single program, and for
program communication?
- is it possible to effectively check this on the synchronous
specification?
In other words, we seek for a theoretical support for a correct move
from synchrony to asynchrony.
- deliverable 4/5.1.1
- (12 months). Characterizing when a single
synchronous program can be regarded as asynchronous without any loss
of semantics with respect to the original specification (this was
last year's deliverable 4.1).
- deliverable 4/5.1.2
- (24 months). Characterizing when a pair of
synchronous programs can be given a fully asynchronous type of
communication without any loss of semantics with respect to the
original specification.
- deliverable 4/5.1.3
- (36 months). Finding how restricted
desynchronisation (e.g., bounded buffering) can be taken advantage
of.
This page has been automatically generated using
Hyperlatex.