Erlang (was: Re: Generics)

Jecel Assumpcao Jr jecel at merlintec.com
Mon Nov 10 19:06:06 UTC 2003


On Sunday 09 November 2003 17:17, Hans Nikolaus Beck wrote:
> Whow could objects be processes ?

Sure, though that depends on what you mean by "object" and what you mean 
be "processes". You might find the old Actor languages (not to be 
confused with the Actors language) an interesting examples of 
object=process.

   http://www.dekorte.com/Actors/Chart.html

> For me objects represent a certain
> state. Their methods are rules how to make a transition to another
> state.

That is one definition. A more popular view around here is that objects 
have some behavior and react to messages that are sent to them. "State" 
is an implementation detail (we could keep a history of all messages 
received instead, for example).

> But processes have no state in that sence that they are a
> chain of infinitesimal transitions. 

That is the first time I have heard anything remotely similar to this. 
It certainly isn't the definition of "process" in Concurrent Sequential 
Processes (CSP), for example. In Unix, Windows, and every other OS I 
know processes do have state (register values, stack, heap, etc...).

> From this theoretical point of
> view, a "process" in a system of objects (i.e. a loop) is not a
> process, only a quick sequence of transitions......
>
> Where is my error in logic ?

Perhaps your logic is ok (I can't tell since the previous paragraph is 
confusing) but your definitions are certainly different.

I have built systems where every single object was a process (see 
http://www.lsi.usp.br/~jecel/tiny.html#rel1), according to popular 
definitions of object and process, and found no problems or 
contradictions.

-- Jecel



More information about the Squeak-dev mailing list