On Tue, Apr 29, 2008 at 09:53:39AM +0800, Ashwin wrote: > > > > > It's surprizing because that call is used quite frequently, e.g. in > > the regression tests, but the entity URI is always NULL which is why this > > was never raised during any of the existing tests... > > I applied and commited a version based on your patch, > > Hi, > Yes, it will be not NULL in a very weird case, somewhat similar to the > one for which there was a fix recently (SVN 3713). Suppose you have an xml > document with an external subset, In the external subset a parameter > entity(say E1) is defined whose replacement text is external using SYSTEM, > Then in the external subset you have another PE (E2) whose replacement text > is E1, in this case entity->URI will not be NULL and would lead to a double > free... > > An extremely weird scenario!!! I don't think anyone would be twisted enough > to use PE's that way....
If you could provide a set of data, i think this is weird enough that this should be added to the test suite to avoid tripping on this later, thanks ! Daniel -- Red Hat Virtualization group http://redhat.com/virtualization/ Daniel Veillard | virtualization library http://libvirt.org/ [EMAIL PROTECTED] | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/ _______________________________________________ xml mailing list, project page http://xmlsoft.org/ [email protected] http://mail.gnome.org/mailman/listinfo/xml
