Ok, I've been looking at the diffs
in wsd2java it was
namespaceMap.put(namespace, packageName);
but in java2wsdl it was
namespaceMap.put(packageName,namespace);
so my unification of the mapping class broke the java2wsdl chain. Now it
makes sense.
I do want the unification to go through, as that lets us have a unified
<mappingset> datatype, and to use mapping files across both tasks.
I also think the wsdl2java order should take priority, as it is what we have
been using in the NStoPkg.properties file; we need to retain that order to
avoid breaking stuff. Dims patches will probably correct java2wsdl, but at
the expense of wsdl2java.
options
-unify the order of ns/package mapping across tasks, though the rationale
for the inversion makes sense when you think about it (one way you map from
ns to package, one way you dont)
-modify mapping and subclasses to have a boolean packageFirst flag which
controls the order
I actually quite like the second option, it gives us unity at the task
level, without making changes down below
----- Original Message -----
From: "Tom Jordahl" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Friday, October 25, 2002 9:58 AM
Subject: RE: cvs commit: xml-axis/java/src/org/apache/axis/wsdl/fromJava
Namespaces.java
>
> Yes, Dims make sure you are not reversing the sense of the mapping form
the command line.
> That would be *VERY* bad.
>
> --
> Tom Jordahl
> Macromedia Server Development
>
>
>
> -----Original Message-----
> From: Steve Loughran [mailto:steve_l@;iseran.com]
> Sent: Friday, October 25, 2002 12:44 PM
> To: [EMAIL PROTECTED]
> Subject: Re: cvs commit: xml-axis/java/src/org/apache/axis/wsdl/fromJava
> Namespaces.java
>
>
>
> ----- Original Message -----
> From: <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Friday, October 25, 2002 9:07 AM
> Subject: cvs commit: xml-axis/java/src/org/apache/axis/wsdl/fromJava
> Namespaces.java
>
>
> > dims 2002/10/25 09:07:49
> >
> > Modified: java/src/org/apache/axis/wsdl Java2WSDL.java
> > java/src/org/apache/axis/wsdl/fromJava Namespaces.java
> > Log:
> > Fix one more "ant clean all-tests" failure.
>
> now I'm really confused.
> This inversion is in code that I didnt touch, which implies the real
defect
> was actually elsewhere in my <mapping> class, which has mixed up key and
> value .
>
>
