Venkata Krishnan wrote:
Hi,
Dealing with the 'appliesTo' attribute in PolicySets has been a subject that
I ended up discussing on the thread
http://www.mail-archive.com/tuscany-dev%40ws.apache.org/msg27768.html. I
have gone forward with a suggestion that Sebastien sounded off on that
thread and have committed the changes under r620307.
First to set the stage....
-----------------------------------
- The PolicySets could be defined in various definitions.xml which are all
read and aggreated for a domain
- Each PolicySet defines an 'appliesTo' attribute which is an xpath that
specifies the sca artifacts to which this policyset may apply.
The problem is about being able to validate the computed or calculated
policysets for a binding / implementation using this 'appliesTo' attribute.
Here is a summary of what's been done.....
---------------------------------------------------------------
- In the contribution read phase, we postpone the reading of composite files
so that all definitions.xml file contents can all be aggregated
- After reading all other kinds of artifacts, we finally read the composite
files in the contribution. The composite xml is first read as a xml
document and for each PolicySet defined for the domain we run the appliesTo
xpath against this xml document. For the nodes returned as result of this
xpath evaluation, we add an attribute named 'applicablePolicySets' whose
value will be the name of the PolicySet in question. So the read composite
will now be modified to include this attribute wherever applicable.
- The modified composite xml is then passed to the STaX processors for the
usual parsing and creation of the the assembly model objects.
- Then during the computing / calculation of PolicySets for a
PolicySetAttachPoint (i.e. a binding or an implementation) the matching
policysets are validated against the list that has been made in the
'applicablePolicySets' attribute of the PolicySetAttachPoint.
As a result of this our samples now don't define their own intents to target
specific policysets for specific references / services. Instead this
targeting is achieved by the proper specification of the 'appliesTo'
attribute in the PolicySet.
Please let me know your thoughts on this.
Thanks
- Venkat
Looks good to me. I am assuming that 'applicablePolicySets' is a
transient and calculated attribute used by Tuscany to flow policySet
info with model elements across the composite reading phases, and not
exposed to application developers.
Were you able to leverage these changes to simplify the bigbank scenario?
--
Jean-Sebastien
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
