Ng Chan Wah wrote:
That's a great idea. But it still requires that the parameter value to be
passed to the service, either by URL or otherwise, be known at the time the
rule was written (using your example, it was known before hand that the
paremeter "user-id" is going to hold the value "staff").
Hm, I see your point. It would probably be possible to allow
placeholders for property values that are already defined in the rule
language (i.e. the IRML message/system/service properties). This would
allow for cases as you mentioned above.
Other opinions on that? Is this a requirement that we should include
in the policy requirements document, and more specifically is this a
requirement for rule languages? I.e. is it required that the rule
language allows
(a) specification/indication of service parameters,
(b) late binding of service parameters, i.e. usage
of placeholders in the rule specification and
binding of actual values to these parameters at
runtime.
It might also be possible to separate service parameter specification
from the rule language and move it into the remote execution
environment (e.g. why should the rule author care about specific
service parameters)? It would than be the responsibility of the remote
execution environment to determine and set the correct parameter
values (for example based on meta-information on the service and
actual values of environment variables or so...). Any views on that?
-Markus