antianthropomorphism

16 views
Skip to first unread message

Douglas Crockford

unread,
Mar 13, 2024, 1:24:59 PMMar 13
to friam

In describing ideal actor systems, I have talked about achieving some level of resilience by stopping wobbling actors and their correspondents. In describing this, I have used words like 'kill', 'murder', and 'suicide'. 'Death before confusion' is a powerful slogan, but it can lead to a very dark and unnecessary place.

So I will now be using words like 'stop', 'halt', and 'cease' instead.

I now look at 'actor' and think that it too may be problematic. We use it as Hewitt did, as  'a thing that acts', but it has much stronger associations with 'a person who acts'. I think this small confusion has made some people suspicious of 'The Actor Model'.

At this stage in the evolution of the model, there are lots of systems that claim to be actor systems that I think are not.

So I am going to stop using 'actor'. Until I figure out the better word, I will be using 'process'. 'Process' is already a term of art, but it comes much closer to describing the communicating distributed objects that I want to build than 'actor' does.

Mark S. Miller

unread,
Mar 13, 2024, 1:33:13 PMMar 13
to fr...@googlegroups.com
Vat?

--
You received this message because you are subscribed to the Google Groups "friam" group.
To unsubscribe from this group and stop receiving emails from it, send an email to friam+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/friam/60fbd9aa-038d-4b20-a048-5e4f57b4535fn%40googlegroups.com.

Tony Arcieri

unread,
Mar 13, 2024, 1:36:21 PMMar 13
to fr...@googlegroups.com
On Wed, Mar 13, 2024 at 11:25 AM Douglas Crockford <dou...@crockford.com> wrote:

Until I figure out the better word, I will be using 'process'. 'Process' is already a term of art, but it comes much closer to describing the communicating distributed objects that I want to build than 'actor' does.

You'd be in good company with Erlang on "process", but it's a term that also has a lot of associated POSIX baggage (a similar situation to "capability" which necessitated the "OCap" rebranding)

--
Tony Arcieri

Alan Karp

unread,
Mar 13, 2024, 1:52:41 PMMar 13
to fr...@googlegroups.com
I think "process" has the right connotations.  In common parlance a process is a unit of failure that has private memory that is not generally available to other processes.  The only downside that occurs to me is that people think of processes as heavyweight things that don't usually send messages to each other.   In that sense, "vat" might be a better term, but you'll need to explain its properties, something using "process" doesn't require.

--------------
Alan Karp


--

Bill Frantz

unread,
Mar 13, 2024, 5:12:36 PMMar 13
to Design
I believe EROS and Capros used process for what KeyKos called a Domain.

Bill

Tristan Slominski

unread,
Mar 14, 2024, 11:05:38 AMMar 14
to friam
In the everything is an actor framing, there's three kinds of actors, two of which don't seem to be as well served by "process". Three kinds I think of are: value actors, function actors, and serial actors. Value actors are essentially constants. Function actors have no memory (like mathematical functions, they always produce the same outputs for the same inputs). Only serial actors have memory which resemble a process.

Douglas Crockford

unread,
Mar 14, 2024, 12:36:20 PMMar 14
to friam
Processes come with inter-process communication. From this perspective, we are applying capability discipline to that communication, allowing it to conveniently and securely cross machine boundaries.

The title of my next talk was going to be "Actors All The Way Up", but that no longer fits. My new tentative title is McLuhanesque, "The Program Is The Process", which isn't nearly as good.

Reply all
Reply to author
Forward
0 new messages