[
https://issues.apache.org/jira/browse/WSCOMMONS-351?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rich Scheuerle updated WSCOMMONS-351:
-------------------------------------
Attachment: patch.txt
Here is the patched code and new test
> OMElement.getChildrenWithName(QName) breaks customers dependent on the old
> semantics
> ------------------------------------------------------------------------------------
>
> Key: WSCOMMONS-351
> URL: https://issues.apache.org/jira/browse/WSCOMMONS-351
> Project: WS-Commons
> Issue Type: Bug
> Components: AXIOM
> Reporter: Rich Scheuerle
> Assignee: Rich Scheuerle
> Attachments: patch.txt
>
>
> Background:
> OMElement.getChildrenWithName(QName searchQName) relies on
> OMChildrenQNameIterator to get the child elements that have a matching qname.
> The OMChildrenQNameIterator had a "loose" interpretation of equality. If
> the child element's local name matched the searchQName local name, then the
> child was returned.
>
> The OMChildrenQNameIterator semantic was changed to a proper, "strict"
> interpretation. (http://svn.apache.org/viewvc?view=rev&revision=522259)
> Problem:
> Some customers built applications that relied on the old "loose"
> interpretation. Many of these customers cannot change their applications
> because they are in a shipped product. When they upgrade to the new Axiom
> levels, their applications fail. (Since Axiom is close to the bottom of the
> stack, it is difficult to diagnose these failures).
> Solution:
> I want to keep the OMChildrenQNameIterator intact. It will continue to
> use the "strict" interpretation.
> In OMElementImpl.getChildrenWithName, I want to use the "strict"
> interpretation if a fully qualified searchQName is provided.
> However, in the case where the searchQName has no namespace, I want to
> have a fallback.
> If there are no child elements that match the unqualified qname, then I
> would like the code to fallback to the old semantic.
>
> I feel that this tactical approach is the best compromise between the
> correct "strict" semantics, but also tolerates customers that relied on the
> old behavior.
> I will provide a patch containing both the new code, trace, and validation
> testcase.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
