[
https://issues.apache.org/jira/browse/WSCOMMONS-111?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Andreas Veithen updated WSCOMMONS-111:
--------------------------------------
Priority: Minor (was: Major)
Fix Version/s: (was: Axiom 1.2.8)
Axiom 1.2.9
The issue with org.apache.axiom.om.impl.traverse.OMChildrenIterator#next() has
been fixed (see WSCOMMONS-372). I also agree that StAXOMBuilder.next() should
not intercept runtime exceptions. Indeed it doesn't make much sense to
intercept a NullPointerException and wrap it in an OMException, since this will
only obscure the root cause of the problem. I changed the exception handling in
StAXOMBuilder.next(), but this causes failures in the Axis2 test suite, so I
will not commit this change for the moment. Since Axis2 has now entered the
release process for version 1.5, I postpone the fix for this Axiom issue to the
next minor release, i.e. 1.2.9. Also, since the primary issue (the behavior of
OMChildrenIterator#next()) has been fixed, I downgrade the priority to minor.
> Careless exception handling needs to be fixed
> ---------------------------------------------
>
> Key: WSCOMMONS-111
> URL: https://issues.apache.org/jira/browse/WSCOMMONS-111
> Project: WS-Commons
> Issue Type: Bug
> Components: AXIOM
> Reporter: Dennis Sosnoski
> Assignee: Andreas Veithen
> Priority: Minor
> Fix For: Axiom 1.2.9
>
>
> org.apache.axiom.om.impl.builder.StAXOMBuilder.next() catches all exceptions
> and rethrows them as OMExceptions (line 199) while other code assumes any
> OMExceptions just mean that the end of the document has been reached (for
> example, org.apache.axiom.om.impl.traverse.next(), line 117, which just sets
> a flag). This means that any coding errors which result in runtime exceptions
> will be masked.
> In the case I've been tracking, an error in
> org.apache.axiom.soap.impl.builder.SOAPBuilderHelper.processAttributes (not
> checking for a null namespace uri) resulted in a NullPointerException, which
> was masked by this code and led to an element being missing from Fault
> details.
> If this code needs to catch particular exceptions these exceptions should be
> individually checked. If this is impractical, at an absolute minimum this
> should check for a RuntimeException subclass and just rethrow it rather than
> masking it as an OMException.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
