[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: continuations in the real world?

Zooko wrote:
> There is a big debate in the world of "peer-to-peer", "web services", et cetera
> about programming style for distributed computation.  One position is "REST",
> which advocates using HTTP GET and PUT for all of your remote invocations.

er.. don't forget POST and DELETE. Also the argument is a little more
subtle than restricting yourself to those four but this isn't the right

> ...  One
> position is normal old RPC.  A third doesn't have a name yet AFAIK, but it is
> non-blocking RPC which results in something like continuation-passing style.

It seems like many in the enterprise software world call queued,
non-blocking message passing just "messaging". Or is this different?

CPS seems to me to be an implementation technique and not something that
is expressed on the wire. (but I admit to being ignorant of E and its
distributed computing model)

Also, REST seems to me to be very compatible with CPS because one
generates documents with URIs that represent the "rest of the
computation." By maintaining a hashtable of URIs to continuations, you
can get the benefit of threading without the locking etc. hassles.

 Paul Prescod