You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jena.apache.org by "Damian Steer (JIRA)" <ji...@apache.org> on 2012/09/19 12:48:07 UTC
[jira] [Comment Edited] (JENA-328) NodeValue makes assumptions
about ClassLoader --- "nasty problem in OSGi environments"
[ https://issues.apache.org/jira/browse/JENA-328?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13458553#comment-13458553 ]
Damian Steer edited comment on JENA-328 at 9/19/12 9:46 PM:
------------------------------------------------------------
Thanks for the report, and suggested fix.
I did look in to DatatypeFactory, since that is a pretty scary warning. My reading is, first, DatatypeFactory is a bit of a mess. Second the idea is to do something like:
String implClass = System.getProperty(
DatatypeFactory.DATATYPEFACTORY_PROPERTY,
DatatypeFactory.DATATYPEFACTORY_IMPLEMENTATION_CLASS);
which addresses the warning (I hope). Will make this change if it sounds ok.
Update:
Looked at the jaxp spec, and it's harder than this. Ugh! Will investigate.
was (Author: shellac):
Thanks for the report, and suggested fix.
I did look in to DatatypeFactory, since that is a pretty scary warning. My reading is, first, DatatypeFactory is a bit of a mess. Second the idea is to do something like:
String implClass = System.getProperty(
DatatypeFactory.DATATYPEFACTORY_PROPERTY,
DatatypeFactory.DATATYPEFACTORY_IMPLEMENTATION_CLASS);
which addresses the warning (I hope). Will make this change if it sounds ok.
> NodeValue makes assumptions about ClassLoader --- "nasty problem in OSGi environments"
> --------------------------------------------------------------------------------------
>
> Key: JENA-328
> URL: https://issues.apache.org/jira/browse/JENA-328
> Project: Apache Jena
> Issue Type: Bug
> Components: ARQ, Jena
> Affects Versions: Jena 2.7.3
> Environment: OSGi: Virgo Server for Apache Tomcat
> Reporter: Martin W.
> Priority: Minor
>
> com.hp.hpl.jena.sparql.expr.NodeValue.java: in lines 175ff:
> A new instance of a DatatypeFactory is created with .newInstance() --- there assumptions are made about the classloader. Such things make it difficult to use Jena in an OSGi-Environment like Virgo.
> Here's my fix: (get the correct classloader first and call newInstance with that classloader as parameter)
> {
> try {
> ClassLoader cl = com.hp.hpl.jena.sparql.expr.NodeValue.class.getClassLoader();
> xmlDatatypeFactory = DatatypeFactory.newInstance(DatatypeFactory.DATATYPEFACTORY_IMPLEMENTATION_CLASS, cl) ; }
> catch (DatatypeConfigurationException ex)
> { throw new ARQInternalErrorException("Can't create a javax.xml DatatypeFactory", ex) ; }
> }
> This fix did it for me. However, the JavaDoc for DatatypeFactory.DATATYPEFACTORY_IMPLEMENTATION_CLASS states "Users should not refer to this field; it is intended only to document a factory implementation detail. "
> NOTE: furthermore I added the original exception ex to:
> throw new ARQInternalErrorException("Can't create a javax.xml DatatypeFactory", ex)
> I think I have seen similar situations in the Jena source code several times --- I would like to see the original exception handed over when new ones are created whenever possible. Using Jena in an OSGi-Environment this would make my life a lot easier.... ;)
> Should i file separate bugs for these things in the future?
> For more information on OSGi and classloaders see also http://stackoverflow.com/questions/1043109/why-cant-jaxb-find-my-jaxb-index-when-running-inside-apache-felix
> Hanno Fietz wrote there "... quite nasty problem in OSGi environments when you're using libraries that aren't designed for OSGi and make assumptions about the classloader they get."
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira