chemistry-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Florent Guillaume ...@nuxeo.com>
Subject Notes from the F2F meeting
Date Tue, 13 Apr 2010 10:30:11 GMT
These is the current status of the discussions at the F2F.

* A close() method is needed on the various session, provider and
implementations of the bindings to be able to release resources
preemptively, in the same way it's done for instance for JDBC
connections. We want clients to be able to follow the try { } finally
{ release stuff } pattern.

* In addition to AtomPub an SOAP, a "Local" binding will be added to
make direct Java calls to the backend implementation without a network
layer. (This can then be used by unit tests to test a backend using
the client API.)

* Server-side SPI
- We'll make the server-side SPI a single "Service" interface extend
the 9 client-side SPI interfaces and add to it some server-specific
methods (deleteObjectOrCancelCheckOut, applyAcl (for AtomPub),
create() without a specific type).
- The CallContext will disappear from the signatures and instead be
passed to the construction of the "ServiceFactory"; each backend can
then hold to the CallContext as it wishes (thread-local, field, etc.).
- The ObjectInfoHolder will not be systematically passed in the
signature but a "objectInfoNeeded" hints will be passed in the
CallContext, and backends will be free to heed it and fill an
ObjectInfo (held by CallContext) or not. The server-side protocol
layer can do the expensive work of calling back to the backend if the
backend did not follow the hints.

Please comment or ask for clarifications if needed.

Thanks,
Florent

-- 
Florent Guillaume, Director of R&D, Nuxeo
Open Source, Java EE based, Enterprise Content Management (ECM)
http://www.nuxeo.com   http://www.nuxeo.org   +33 1 40 33 79 87

Mime
View raw message