@Steve Ebersole The issue already affected older versions of Hibernate (pre-4.3.0), in which JarVisitorFactory was still present. However, the issue is not specific to JarVisitorFactory - it still present in 4.3.8.Final.

I have tested the pull requests only on JBoss 4.3.0, not with other application servers. At least for the time being I don't have resources to do large-scale testing on multiple application servers.

Since the pull requests should first do what 4.3.8.Final does (resolve the file location using the JVM working directory) and only attempt relative paths if the 4.3.8.Final-approach fails, the regression risk should be low. But as always, coding is full of surprises.