axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Russell Butek" <bu...@us.ibm.com>
Subject Re: WSDL2Java: JavaFaultWriter problem with imports
Date Mon, 17 Dec 2001 17:04:29 GMT
The problem is simply one of not regenerating the fault class every time a
fault is seen.  The exception generated is named, not after the fault in
the operation, but after the message name that that fault refers to.  So
there is the potential for any number of operations on any number of
portTypes to refer to the same message and hence generate the same
exception.  If you're looking at a portType, you can collect all the faults
that use the same message and generate that fault once.  But other
portTypes might also have faults that use the same message.  By generating
faults at a higher level than portType, we can look at all faults in all
portTypes to prevent the exception from being generated more than once.

By the way, the relative-directory problem on imports appears to be in
WSDL2Java, so it's something I should be able to fix today, along with the
missing fault problem.

Russell Butek
butek@us.ibm.com


Thomas Sandholm <sandholm@mcs.anl.gov> on 12/17/2001 10:49:41 AM

Please respond to axis-dev@xml.apache.org

To:   axis-dev@xml.apache.org
cc:
Subject:  Re: WSDL2Java: JavaFaultWriter problem with imports



At 10:24 AM 12/17/2001 -0600, Russell Butek wrote:

>The problem with generating faults - and correct me if I'm wrong, Tom - is
>that we need info from all portTypes in order to find all faults and make
>sure we don't try to generate duplicates.  JavaPortTypeWriter only gives
us
>info on a particular portType.  But I think it would be sufficient for the
>logic to look at all portTypes in all Definitions, primary and imported.
>I'll look into it.

Well, I might be missing something here in how Axis is designed internally
(or some Java restrictions), but theoretically I see no big difference in
fault parts, input parts, and output parts of a portType. A fault part type
may be reused in many different portType definitions, likewise input and
output part types may also be reused among portTypes. In my tests I have
both types of reuse, and the input, output part type kind of reuse seems to
work smoothly with imports. So, I guess what I'm saying is that, if
possible, it would be nice if these portType parts could be handled
consistently.

/Thomas




Mime
View raw message