[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [sca-policy] ISSUE 46: How to configure policySets
Hi Ashok, Thanks for the clarifications. Regarding the symmetry part - I think one major unsymmetry we recently (forgot the issue#) introduced is that while intents can be applied at any level, policySet is now applicable only to the binding node. Therefore syntactically it would be easier to define policySet as an element by expanding upon the definition of the binding element. OTOH, supporting 'requires' as an element would require adding an extra optional element at all levels in the SCDL hierarchy, which may make the SCDL a bit unwieldy. We will also have to expand upon the inheritance and conflict resolution rules if intents were to be an element, since the current rules are in terms of only a 'list of QNames' as value for the 'requires' attribute. Moreover, I am still not sure about the rationale for making 'requires' an element. I think 'requires' should stay as an attribute unless a clear case is made for promoting it to an element. -- Sanjay > -----Original Message----- > From: ashok malhotra [mailto:ashok.malhotra@oracle.com] > Sent: Wednesday, Jan 30, 2008 5:33 AM > To: Patil, Sanjay > Cc: OASIS Policy > Subject: Re: [sca-policy] ISSUE 46: How to configure policySets > > Hi Sanjay: > > Yes, we do not want to standardize on a configuration model. > > No, there is no attempt to change or reduce the intent functionality. > > I made "requires" a child element merely for symmetry. If > folks want > to retain it as an attribute, > I'm fine with that. > > Ashok > > Patil, Sanjay wrote: > > >[Can we have an issue number assigned asap, please] > > > >Some comments inline ... > > > > > > > >>-----Original Message----- > >>From: ashok malhotra [mailto:ashok.malhotra@oracle.com] > >>Sent: Tuesday, Jan 29, 2008 6:29 AM > >>To: OASIS Policy > >>Subject: [sca-policy] NEW ISSUE: How to configure policySets > >> > >>TARGET: SCA Policy Framework > >> > >>DESCRIPTION : > >> > >>The current SCA Policy Framework supports the specification > >>of intents > >>and policySets on SCDL elements using attributes. > >>For example: > >> > >><service> or <reference> > >> > >><binding name = "xs:string" > >> > >>policySet="xs:QName"? requires="="listOfQNames"? /> > >> > >></service> or </reference> > >> > >> > >>There are a couple of problems with this approach. > >> > >> > >>1. Attributes are not extensible. If the intents and > policySets were > >>specified as child elements, this would permit > >>extensibility. In the f2f meeting last week the argument was > >>made that > >>some extensibility was required to allow > >>runtime extension of policies with information like the > >>location of the > >>key store or the type of certificate required. > >> > >> > > > >I am guessing that you want to support binding configuration > by allowing > >extensibility of policySets but not standardize the > configuration model > >itself. Am I right? > > > > > > > > > >>2. If the 'requires' attribute is used in conjunction with > >>a policySet > >>attribute it's not clear what the semantics are. > >>At one time the intents in the 'requires' list were used to > >> configure > >>the policySets but that meaning seems to have been lost. > >> > >> > > > >I think we would still want to keep the intents mechanism > such that the > >developers can express abstract requirements and deployers > can validate > >any selected bindings for the concerned wire, right? > > > > > > > >>PROPOSAL: > >> > >>For the reasons cited above, I suggest that policySets and > >>intents be > >>associated with SCDL elements using child elements. > >>For example: > >> > >><service> or <reference> > >> > >><binding name = "xs:string" > >> > >><policySet name="xs:QName" select="xs:string"*> * > >> <requires name ="xs:QName"/> * > >> > >></service> or </reference> > >> > >> > > > >I understand why we might want to have policySet as an > element. I am not > >sure why we would want to make 'requires' as an element (vs. an > >attribute)? > > > >-- Sanjay > > > > > > > >>Each "policySet" element would name a single policySet. > The "select" > >>attribute could contain a list of xs:strings which would > >>name qualified intents and customize the use of the policySet by > >>selecting specific options from the intentMaps within the policySet. > >> > >>For symmetry, the "requires" element also names a single > intent but > >>this is not necessary. It could be changed to take a > >>list of intent names as long as there was no need to customize the > >>intents. > >> > >>-- > >>All the best, Ashok > >> > >>------------------------------------------------------------ > --------- > >>To unsubscribe from this mail list, you must leave the OASIS TC that > >>generates this mail. You may a link to this group and all > >>your TCs in OASIS > >>at: > >>https://www.oasis-open.org/apps/org/workgroup/portal/my_workgr > >>oups.php > >> > >> > >> > >> > > > -- > All the best, Ashok > > --------------------------------------------------------------------- > To unsubscribe from this mail list, you must leave the OASIS TC that > generates this mail. You may a link to this group and all > your TCs in OASIS > at: > https://www.oasis-open.org/apps/org/workgroup/portal/my_workgr > oups.php > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]