'Re: [Neethi] More updates, much easier to port from 2 -> 3'

[prev in list] [next in list] [prev in thread] [next in thread] 

List:       xmlrpc-dev
Subject:    Re: [Neethi] More updates, much easier to port from 2 -> 3
From:       Daniel Kulp <dkulp () apache ! org>
Date:       2011-02-21 16:32:51
Message-ID: 201102211132.51674.dkulp () apache ! org
[Download RAW message or body]


I've added some test cases for the stuff below and fixed the code.

Thanks for the feedback!

Dan


On Saturday 19 February 2011 2:32:07 PM Daniel Kulp wrote:
> Andreas,
> 
> Major thanks for looking at this.   It's a big help since you obviously
> have found some issues that need to get fixed before the release.
> 
> #2 is definitely a problem I overlooked.   I'll get that fixed on Monday.
> 
> #1 is a bigger....
> 
> On Saturday 19 February 2011 10:30:59 AM Andreas Veithen wrote:
> > I took over the task of porting Sandesha2 to Neethi 3. That turned out
> > to be trickier than Axis2 and Rampart. I have an issue with the
> > 
> > following code in XMLPrimitiveAssertionBuilder:
> >         if (count == 0) {
> >         
> >             return newPrimitiveAssertion(element, atts.isEmpty() ? null :
> > atts); } else if (policyCount == 1 && count == 1) {
> > 
> >             Policy policy =
> >             factory.getPolicyEngine().getPolicy(policyEl); return
> >             newPolicyContainingAssertion(element,
> > 
> > atts.isEmpty() ? null : atts, policy);
> > 
> >         }
> >         return new XmlPrimitiveAssertion(element);
> > 
> > There are two problems with that logic:
> > 
> > 1. For assertions that have an optional child element, this logic
> > implies that the builder returns two fairly different types of objects
> > depending on the actual content of the assertion: a PrimitiveAssertion
> > if there is no child element and an XmlPrimitiveAssertion if there is
> > one. Note that in contrast to what the names may suggest,
> > XmlPrimitiveAssertion doesn't extend PrimitiveAssertion and they only
> > have the Assertion interface in common. That doesn't make sense to me;
> > two assertions differing only by the number of child elements should
> > not be represented by two completely different classes.
> 
> Well, I debated removing the XmlPrimitiveAssertion class entirely as it
> really cannot be used for much other than parsing.   It doesn't support
> normalization properly nor can it really handle intersections.   It could
> definitely extend the PrimitiveAssertion, but the
> normalization/intersetion issues would remain.
> 
> > Of course one can argue that PrimitiveAssertion and
> > XmlPrimitiveAssertion should be considered as opaque, i.e. that they
> > are used only to ensure round-trippability, but that client code
> > should not attempt to parse the content of these types of assertions.
> > In fact one can consider that if the client code needs to parse an
> > assertion, it should have an appropriate assertion builder.
> 
> Well, I generatlly consider PrimitiveAssertion and the
> PolicyContainingPrimitiveAssertion as more or less possible base classes
> that customs could subclass.  I've done that in CXF.
> 
> > That is actually the point of view I took when porting Sandesha2.
> > Previously it attempted to parse the content of
> > XmlPrim(i)tiveAssertion objects, and now I've changed the code so that
> > there is an assertion builder for each type of assertion. However,
> > this causes a huge increase of the volume of code necessary to parse
> > assertions.
> > 
> > In addition, if one considers PrimitiveAssertion and
> > XmlPrimitiveAssertion as opaque, then the logical consequence would be
> > that these classes should not even have public methods allowing to
> > retrieve the content.
> 
> Well, one of the things I'm trying to do is make it so a "generic" policy
> editor (maybe a GUI or something), can parse/process policies in a very
> generic way without the custom parsers and such.   Thus, Neethi should at
> least be able to parse/process everything in a generic way without all the
> custom builders and such.   Thus, if info is being lost, that is a problem.
> Thanks for looking at it and finding some of those issues.   I'll look at
> them on Monday.
> 
> Obviously, runtimes that need more complete control over the parsing and
> types should provide custom builders and types and such.
> 
> Dan
> 
> > 2. The code fragment shown above neglects the existence of assertions
> > that have no child elements, but that contain character data. Not sure
> > if this kind of assertions conform to the specs and/or best practices,
> > but they are used in Sandesha2. Example:
> > 
> > <sandesha2:MaximumRetransmissionCount>10</sandesha2:MaximumRetransmission
> > Co unt>
> > 
> > Currently, the code will create a PrimitiveAssertion and drop the
> > content of the assertion (only attributes are stored). That issue is
> > more serious than the first one because it breaks round-trippability.
> > 
> > Andreas
> > 
> > On Fri, Feb 18, 2011 at 17:18, Daniel Kulp <dkulp@apache.org> wrote:
> > > On Friday 18 February 2011 12:12:25 AM Afkham Azeez wrote:
> > >> Dan,
> > >> We will need to organize a hackathon to migrate Axis2 to use the new
> > >> Neethi APIs. Will you be able to assist during this?
> > > 
> > > I'm not really sure you would need a Hackathon for it.   With the
> > > latest updates, it will probably take a single person longer to run
> > > the tests afterword than it would take to update the code.
> > > 
> > > I just created RAMPART-322 and attached a patch that completely updates
> > > it to Neethi3.   Took about 30 minutes of work to do, and that's mostly
> > > cause I hit a bug in the new Neethi code that needed to get fixed
> > > first.
> > > 
> > > Yes, if you wanted to update all of Axis/Rampart/etc... to not use the
> > > static PolicyEngine, that would require a lot more work.  However, that
> > > isn't required.
> > > 
> > > Dan
> > > 
> > >> Azeez
> > >> 
> > >> On Fri, Feb 18, 2011 at 10:04 AM, Daniel Kulp <dkulp@apache.org> wrote:
> > >> > I just committed a bunch more updates to Neethi that cleanup a few
> > >> > things, but
> > >> > also will end up making it much easier to port from Neeth 2 to 3.
> > >> > 
> > >> > 1) I renamed PolicyEngine to PolicyBuilder.  It really is just a
> > >> > Policy builder.   CXF and Axis (and others) really have the "engine"
> > >> > for asserting the policies and such, thus the rename kind of makes
> > >> > sense.
> > >> > 
> > >> > 2) With that renamed, I re-created the PolicyEngine with all the
> > >> > static methods.  It just wrappers a static instance of
> > >> > PolicyBuilder.
> > >> > 
> > >> >    I'd LIKE to
> > >> > 
> > >> > mark PolicyEngine as Deprecated as it's definitely better to not use
> > >> > the statics, but I'm also OK just leaving it.
> > >> > 
> > >> > 3) I also changed the return of getAlternatives to match what the
> > >> > javadoc said
> > >> > which also matches what Axis is casting it to.  (simplified some CXF
> > >> > code as
> > >> > well)
> > >> > 
> > >> > 
> > >> > The result is that it's now pretty easy to update from 2.x to 3.x.
> > >> > The changes required really are just:
> > >> > 
> > >> > 1) Anything that implements AssertionBuilder will neet to be updated
> > >> > to implement AssertionBuilder<OMElement>.   Fairly simple.
> > >> > 
> > >> > 2) All policies will need to have a "boolean isIgnorable()" method
> > >> > added.
> > >> > 
> > >> >  You
> > >> > 
> > >> > can just add a default method that returns false to keep the old
> > >> > behavior. However, you could also update the builders to record the
> > >> > ignorable flag properly so the intersections would work.
> > >> > 
> > >> > I've included a patch below that would update the axis2 trunk.
> > >> > Obviously, things like Rampart would be more involved, but it does
> > >> > show how simple it is.
> > >> > 
> > >> > Once again, I'd appreciate anyone else looking at it and any
> > >> > thoughts folks may have.
> > >> > 
> > >> > --
> > >> > Daniel Kulp
> > >> > dkulp@apache.org
> > >> > http://dankulp.com/blog
> > >> > 
> > >> > 
> > >> > 
> > >> > 
> > >> > 
> > >> > Index: modules/parent/pom.xml
> > >> > ===================================================================
> > >> > --- modules/parent/pom.xml      (revision 1071847)
> > >> > +++ modules/parent/pom.xml      (working copy)
> > >> > @@ -69,7 +69,7 @@
> > >> > 
> > >> >        <!-- Tracking SNAPSHOT(s) of a few projects -->
> > >> >        
> > >> >         <axiom.version>1.2.12-SNAPSHOT</axiom.version>
> > >> >         <!--TODO: fix the compatibility issues with the Neethi trunk
> > >> > 
> > >> > and
> > >> > 
> > >> > bring back to 3.0.0-SNAPSHOT-->
> > >> > -        <neethi.version>2.0.4</neethi.version>
> > >> > +        <neethi.version>3.0.0-SNAPSHOT</neethi.version>
> > >> > 
> > >> >         <woden.version>1.0-SNAPSHOT</woden.version>
> > >> >        
> > >> >        <!-- Use released versions for these projects -->
> > >> > 
> > >> > Index:
> > >> > 
> > >> > modules/metadata/src/org/apache/axis2/jaxws/description/builder/conv
> > >> > er ter /JavaMethodsToMDCConverter.java
> > >> > ===================================================================
> > >> > ---
> > >> > 
> > >> > modules/metadata/src/org/apache/axis2/jaxws/description/builder/conv
> > >> > er ter /JavaMethodsToMDCConverter.java (revision 1071847)
> > >> > +++
> > >> > 
> > >> > modules/metadata/src/org/apache/axis2/jaxws/description/builder/conv
> > >> > er ter /JavaMethodsToMDCConverter.java (working copy)
> > >> > @@ -28,7 +28,6 @@
> > >> > 
> > >> >  import org.apache.axis2.jaxws.description.builder.WebEndpointAnnot;
> > >> >  import org.apache.axis2.jaxws.description.builder.WebMethodAnnot;
> > >> >  import org.apache.axis2.jaxws.description.builder.WebResultAnnot;
> > >> > 
> > >> > -import sun.reflect.generics.reflectiveObjects.GenericArrayTypeImpl;
> > >> > 
> > >> >  import javax.jws.Oneway;
> > >> >  import javax.jws.WebMethod;
> > >> > 
> > >> > @@ -39,6 +38,7 @@
> > >> > 
> > >> >  import javax.xml.ws.ResponseWrapper;
> > >> >  import javax.xml.ws.WebEndpoint;
> > >> >  import java.lang.reflect.Constructor;
> > >> > 
> > >> > +import java.lang.reflect.GenericArrayType;
> > >> > 
> > >> >  import java.lang.reflect.Method;
> > >> >  import java.lang.reflect.Modifier;
> > >> >  import java.lang.reflect.ParameterizedType;
> > >> > 
> > >> > @@ -347,7 +347,7 @@
> > >> > 
> > >> >             mdc.setReturnType(fullType);
> > >> >         
> > >> >         } else if (type instanceof Class) {
> > >> >         
> > >> >             mdc.setReturnType(((Class)type).getName());
> > >> > 
> > >> > -        } else if (type instanceof GenericArrayTypeImpl) {
> > >> > +        } else if (type instanceof GenericArrayType) {
> > >> > 
> > >> >             mdc.setReturnType(type.getClass().getName());
> > >> >         
> > >> >         }
> > >> >     
> > >> >     }
> > >> > 
> > >> > Index:
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/model/MTOMAssertion.j
> > >> > av a
> > >> > ===================================================================
> > >> > ---
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/model/MTOMAssertion.j
> > >> > av a (revision 1071847)
> > >> > +++
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/model/MTOMAssertion.j
> > >> > av a (working copy)
> > >> > @@ -52,5 +52,10 @@
> > >> > 
> > >> >         this.optional = isOptional;
> > >> >     
> > >> >     }
> > >> > 
> > >> > +
> > >> > +    public boolean isIgnorable() {
> > >> > +        return false;
> > >> > +    }
> > >> > +
> > >> > 
> > >> >  }
> > >> > 
> > >> > \ No newline at end of file
> > >> > Index:
> > >> > 
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/builders/MTOM10Assert
> > >> > io nBu ilder.java
> > >> > ===================================================================
> > >> > ---
> > >> > 
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/builders/MTOM10Assert
> > >> > io nBu ilder.java (revision 1071847)
> > >> > +++
> > >> > 
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/builders/MTOM10Assert
> > >> > io nBu ilder.java (working copy)
> > >> > @@ -43,7 +43,7 @@
> > >> > 
> > >> >  * The builder will be picked by the
> > >> >  * "org.apache.neethi.AssertionBuilderFactory".
> > >> >  */
> > >> > 
> > >> > -public class MTOM10AssertionBuilder implements AssertionBuilder {
> > >> > +public class MTOM10AssertionBuilder implements
> > >> > AssertionBuilder<OMElement> {
> > >> > 
> > >> >     private static Log log =
> > >> > 
> > >> > LogFactory.getLog(MTOM10AssertionBuilder.class);
> > >> > 
> > >> > Index:
> > >> > 
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/builders/MTOM11Assert
> > >> > io nBu ilder.java
> > >> > ===================================================================
> > >> > ---
> > >> > 
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/builders/MTOM11Assert
> > >> > io nBu ilder.java (revision 1071847)
> > >> > +++
> > >> > 
> > >> > modules/mtompolicy/src/org/apache/axis2/policy/builders/MTOM11Assert
> > >> > io nBu ilder.java (working copy)
> > >> > @@ -39,7 +39,7 @@
> > >> > 
> > >> >  * "org.apache.neethi.AssertionBuilderFactory".
> > >> >  */
> > >> > 
> > >> > -public class MTOM11AssertionBuilder implements AssertionBuilder{
> > >> > +public class MTOM11AssertionBuilder implements
> > >> > AssertionBuilder<OMElement> {
> > >> > 
> > >> >     private static Log log =
> > >> > 
> > >> > LogFactory.getLog(MTOM10AssertionBuilder.class);
> > >> > 
> > >> > 
> > >> > --------------------------------------------------------------------
> > >> > - To unsubscribe, e-mail: dev-unsubscribe@ws.apache.org
> > >> > For additional commands, e-mail: dev-help@ws.apache.org
> > > 
> > > --
> > > Daniel Kulp
> > > dkulp@apache.org
> > > http://dankulp.com/blog
> > > 
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: dev-unsubscribe@ws.apache.org
> > > For additional commands, e-mail: dev-help@ws.apache.org
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@ws.apache.org
> > For additional commands, e-mail: dev-help@ws.apache.org

-- 
Daniel Kulp
dkulp@apache.org
http://dankulp.com/blog
Talend - http://www.talend.com

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@ws.apache.org
For additional commands, e-mail: dev-help@ws.apache.org

[prev in list] [next in list] [prev in thread] [next in thread]
Configure | About | News | Add a list | Sponsored by KoreLogic