----- Original Message -----
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, June 27, 2002 3:54 PM
Subject: cvs commit: xml-axis/java/src/org/apache/axis/wsdl/fromJava
Types.java
> stevel 2002/06/27 15:54:55
>
> Modified: java/src/org/apache/axis/wsdl/fromJava Types.java
> Log:
> costin's null patch with some bracing round clauses thrown in.
> public QName writePartType(Class type, javax.xml.namespace.QName
qname) throws Exception {
> - if (type.getName().equals("void"))
> + if( type==null ) {
> + return null;
> + }
> + if (type.getName().equals("void")) {
> return null;
> + }
Costin, this is the only one of your null pointer tests that I have
committed, and thinking about it I may revert it if pushed.
1. the one you had on class.forName is subsumed by the ClassUtils stuff
(remember, diffs against CVS_HEAD)
2. The other one was too much symptom fixing, not problem fixing :
--- src/org/apache/axis/description/ServiceDesc.java 4 Jun 2002
20:34:02 -0000 1.24
+++ src/org/apache/axis/description/ServiceDesc.java 27 Jun 2002
20:08:38 -0000
@@ -676,7 +676,8 @@
createOperationsForName(implClass, methodName);
// Note that we never have to look at this method name again.
- completedNames.add(methodName);
+ if (completedNames != null)
+ completedNames.add(methodName);
}
if something is going wrong here, we shouldnt just skip it. Likewise, I fear
the writePartType patch is also hiding another problem, which is why I am
sorely tempted to revert it. If anyone else agrees with me, I will pull it.
All the logging stuff is in there...I think it would be good to have a
constants file that defined all the special logging channels for the
following three reasons.
1. is easy to javadoc
2. less chance of typos
3. easy to rename
I'll leave that as an exercise for you.
