Lists Home |
Date Index |
Paul Prescod wrote:
> "Simon St.Laurent" wrote:
> > I'd love to see an "XMLchucker" protocol that just opens a port, sends
> > the info, and maybe replies with a checksum or an error. No more.
> It'll take about ten minutes to write the RFC for that. But your
> intermediaries will have no idea what is going on and won't be able to help
Unless the nodes which act upon your marked-up message are implementations of
particular processing expertise: capable, that is, of doing useful things
with your message which you cannot do yourself and may well not even
understand in much detail what is to be done. Isn't this intense
specialization of processes, which can then be harnessed case-by-case into a
useful pipeline of operations, really the point of the "web services"
philosophy and indeed of distributed processing generally? You address a
message to a node because you believe its expertise is required at that point
in a chain of process. Or, alternatively, you publish a message, which is then
acted upon by an interested node whose very existence you knew nothing of, let
alone which verbs it implements, or how. Isn't the point of such distributed
processing precisely that it is *not* sub-contracting? This is not a case
where a build-vs-buy analysis indicates that another node can achieve your
expected result more cheaply or quickly than you can do it yourself, so you
consign some raw material to that node for a particular step of processing.
Distributed processing is specifically about what you cannot do yourself--nor
may even understand can be done--rather than what you want to delegate, but
control, from a distance. That is why RPC, by its very nature, sets out to
achieve the converse of distributed cooperation.