[MOBY-dev] RFC #1941: INB Asynchronous Service Call Proposal - discussion summary (so far)

[Johan Karlsson]

Hello everyone,

It is very encouraging to get such a reaction to a RFC. It clearly 
demonstrates that the proposal is needed. The result of all these 
discussion and comments will be a much-needed mechanism to call 
long-running services (something we are eager to implement in our system).

However, let me first point out that the RFC is a collaborative result 
from the entire INB.

Martin suggested me to summarize what has been discussed so far and this 
is what I have come up with (really, it is not an easy task...). Please, 
if I misinterpret "agreement", I apologize and ask that you let me know. 
The motivation for the summary is to see if we are reaching some form of 
consensus. It does not mean that discussion on the "agreed" points is 
over, anyone with comments are very welcome.

"agreed?":
========

* Three additional methods to represent starting an asynchronous 
session, polling for status and getting the result.
* It should be allowed to retrieve results or status for specific 
mobyData inputs ("jobs"), although all clients must not support this (if 
they do not want)
* XML from LSAE should be used to carry status information (somehow)
* More specifics needed about which mobyExceptions for async-related 
fails/warnings
* Information about if a service can be run asynchronously should be 
stored in the registry.

Still debated or unresolved:
=========

* Using mobyStatus vs. using status objects in the ontology (actually 
this is the most debated question). (Martin)
* How clients determine if a service can be called asynchronously (Martin)
* Names of the methods, use instead verbs to better describe the 
methods. (Pieter Neerincx)
* Alternate solution, WSRF (this one really would mean a completely new 
RFC) (Tom Oinn)
* _result cleans the result vs a specific _clean (Martin)
* session_id (Martin)

Also, there has been several (justified) requests for rewrites (unclear 
sentences, better documented predicate etc) to clarify what is meant. We 
are working on this.

Additional comments coming in answers to your specific emails.

So far, many and valuable comments from the community but we want to 
hear from more people what they think.

Kind regards,

Johan