Hi,
I'm asking for a little advise about practices here. I'm new with web
services and we're planning to expose our internal SOA with a web services
interface.
We already have input and output beans corresponding to input and output
messages for our internal invoker. But we need input and output web service
messages to be different than the one's we already have. Thus, a mapping
between web service messages beans and internal beans should be made. The
thing is, we don't want to have 2 beans for input and 2 beans for output
that looks practicaly the same (since a sysmtematic mapping could transform
a web service message bean into a internal bean and viceversa) plus a method
for each service we already have, for each and every single service (and
there are a lot).
The solution I came up is working with MESSAGES services in orden to have a
centraliced entry point which should parse the xml by hand and instance an
already-existing bean according to the service the third party is invoking.
That will mean writing ONLY this unique piece of code.
The question is, am I doing right? or should I use wrapped document services
and make special serializers/deserializersm, or making use of some
data-binding framework? (taking into account I only want 1 entry point that
act as a kind of "business delegate" and "facade", so that exposing an
internal service as web service would only imply a parametric change in the
system metadata, for example, setting the "expose as web service" flag for a
specific service on).
Thank you very much in advance for the advise.

Reply via email to