On Fri, 15 Feb 2002, Alan Kay wrote:

> Bijan, as you say, Erlang programming is very similar to Smalltalk-72 
> programming, since message receiving is implemented in much the same 
> way (and both are derived from a kind of coroutine structure on 
> something like a closure).

Hmm. The part the felt similar to me is the parsing of the messages in a
kind of case structure. AFAICT, Smalltalk-72 used synchonous calls (at
least, you return to the sender, just like in Smalltalk-80). Whereas, in
Erlang, the only way to get stuff back to the sender is for the sender to
include it's id in the message and for the receiver to explicitly send the
message back.

I was reflecting on the problems you discussed at various points of each
class inventing syntax in Smalltalk-72, and wondered why it doesn't seem
to be a problem in Erlang. The answer seems partially to lie in the fact
that Erlang isn't as systematically message oriented. Though concurrancy
is built in and well integrated, you can do a fair bit in Erlang with
purely sequential code (i.e., there's a reasonable functional programming
langauge in there, next to the message passing stuff). Plus, you build
messages fairly explicitly, rather than casting all syntax into the
message passing mold.

That being said, I don't quite get all of the Smalltalk-72 model, though
I'm diligently (for me) perusing the manual :)

I'll also note that SOAP, the non-rpc bits, are *very* message
oriented. The W3C use case document recently put out has some nice
examples of "thinking in the inbetween, at the messages" stuff that you
sometimes talk of. Alas, I think this will go past many folks (web
services, faugh!).

Cheers,
Bijan Parsia.