[
Lists Home |
Date Index |
Thread Index
]
"Bullard, Claude L (Len)" wrote:
> Interoperability can be achieved by autonomous applications that inquire of a source
> what types it can produce and being able to recognize those types. There is no free
> lunch here. Discovery-based systems will work reasonably well. The web
> functions because the HTTP operations are fixed and their semantics are well
> understood, and because one dominant type (HTML) rules the majority of transactions.
Absolutely agreed. I do try to extend this point, however, by noticing that the most
accurate and effective discovery about a process is in the examination of its output.
You see precisely what you are getting, not only in the instance content, but in its
manifest structure. Furthermore, these outputs as XML, or even as HTML, are documents
and can therefore be published at URLs, allow the whole system of interprocess
communication to be managed, as the RESTies advocate, with the general-purpose tools of
http.
Respectfully,
Walter Perry
|