On Tue, Jan 14, 2003 at 06:08:03AM -0500, Daniel Veillard wrote:
On Tue, Jan 14, 2003 at 10:47:28AM +0100, Stéphane Bidoul wrote: It applies cleanly, it looks sound overall, but it generate memory
Actually, the more I look at it the less I agree with the code. libxml_xmlParserCtxtPtrWrap() allocate an empty xmlParserCtxtPyCtxt each time the parser context reference is passed back to the Python level. Plus it doesn't not even check if ctxt->_private existed or not. I think that's wrong. The ctxt->_private should be set *only* when the Python routine for associating private handlers to the context is called after checking that such a block doesn't exist already. I see no reason why libxml_xmlParserCtxtPtrWrap() should be modified. I fixed this, integrated ctxterror.py in the python/tests regressions resulting in the enclosed patch, which was then commited to CVS thanks, Daniel -- Daniel Veillard | Red Hat Network https://rhn.redhat.com/ veillard redhat com | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/
Attachment:
patch
Description: Text document