Actor languages [was [E-Lang] E FAQ]

[Ken Kahn]

From: "Marc Stiegler" <marcs@skyhunter.com>

> I observer philosophically that the failure of Actors is now easy to
> understand: no one ever actually implemented it! This is even more fatal
as
> a  language introduction strategy than is poor marketing, like Smalltalk
:-)
> Markm, it looks to me like you are making a dramatic contribution with E
> after all, simply completing an actual working language is a fairly big
leap
> forward (well, some of us consider it a big leap forward, I realize that
if
> I were properly theoretical, actual implementation wouldn't count as
> a worthy effort :-)
>

Soon after becoming a graduate student in 1973 I took a course from Carl
Hewitt where we were assigned different pieces of the implementation of
Plasma.  My recollection of that code is that it was the cleanest code I had
ever seen but also ran very slowly. My memory of it was that it ran examples
in papers but no one built anything serious in it.

In 1981 I was building a language I called Uniform (based upon semantic
unification) and decided to use Henry Lieberman's Act 1. Act 1 was less of a
toy implementation than Plasma so my implementation of Uniform ran but was
itself a toy.

For what it is worth my recollection is that Plasma was a few orders of
magnitude slower than compiled Lisp while Act 1 was about one or two orders
of magnitude.

I never used Act 2. (By the way, around 1982 I implemented Intermission
which was an actor language in Prolog - another toy as the name suggested.)

I seem to remember hearing about a more serious implementation of actors in
France during the 80s. (Maybe Henry who I've cc-ed remembers more.) Also Aki
Yonezawa (another student of Carl's) has been active making actor-like
languages.
http://web.yl.is.s.u-tokyo.ac.jp/members/yonezawa/home.html

Best,

-ken kahn ( www.toontalk.com )