>> I can say that flying in an aircraft controlled by Java is
>> probably safer than one controlled by C, until garbage
>> collection kicks in just when you need reverse thrust.
>
>Though I still would not want my life entrusted to Java, it is worth
>pointing out that I read a blurb (can't remember the company name) about a
>JVM that allegedly "guarantees deterministic performance."


I've been lurking on this thread for a while, but I guess I should put in my
$.02 ....

"Vanilla" Java clearly has a number of problems with respect to determinstic
real-time performance: the absence of stack-based objects, the need to
account for gc, the incomplete specification of how priorities affect
scheduling, the absence of provisions for avoiding priority inversion, the
absence of an absolute "sleep", the relatively limited set of facilities for
asynchronous event handling, and several others.  Nevertheless the
language's very high degree of portability and safety (more than Ada's in
fact) have sparked interest from the real-time community.  There was a
requirements-gathering effort last year led by NIST, involving major
stakeholders from both the real-time and Java communities.  Since early in
1999 two efforts have been underway at addressing Java's shortcomings in the
real-time arena: one under the auspices of Sun's Java Community Process, and
the other one (produced by the "J-Consortium") independent of Sun.  There
are no good technical reasons why there should be two such efforts, but
various companies' business interests have led them into one camp or the
other, or into both.   Both groups released their specs at the Embedded
Systems Conference in San Jose last month.  If you want more information,
check their web sites: http://www.rtj.org (the Real-Time for Java Experts
Group under Sun's JCP) or http://www.j-consortium.org (for the
J-Consortium's spec).

Ben Brosgol
[log in to unmask]