DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT <http://nagoya.apache.org/bugzilla/show_bug.cgi?id=7164>. ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND INSERTED IN THE BUG DATABASE.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=7164 DOMParser with a DTD leak! ------- Additional Comments From [EMAIL PROTECTED] 2002-03-25 20:09 ------- Jean, Though your patch seem eliminate the momory leak to this particular bug, but overall, I am afraid it is not the way it is supposed to be. The algo for DOMParser::endAttrList is like this, for any incoming element (ElemDecl in the code) with default attribute, create a temporary element (elem in the code), with default attribute list set, from the element pool (owned by the documentType), then loop through the default attribute list of the incoming element, and insert each attribute to the tempoary element and once finished, insert this temporary element (the elem) into the element pool. The leak occurs if, the temporary element's default attribute list has a same attribute name which exists in the incoming element's default attribute list, it will be overwritten by the new one from the incoming elements', and left unattached to any thing, becoming dangling and therefore, not released properly. To solve this, my patch will capture this kind attribute and delete it if necessary. so instead of simply do elem->setAttributeNode(insertAttr); we check and delete: AttrImpl * previousAttr = elem->setAttributeNode(insertAttr); if ( previousAttr != 0 && previousAttr->nodeRefCount ==0) NodeImpl::deleteIf(previousAttr); The reason why your patch does not have a leak, is, your element will be put into entity pool, not element pool, so any temp element created from element poool natually does not have an initial default attribute list, and there fore no leak occurs, and you happen to delete the element from entity pool if it is already there. I'd suggest you run samples from the latest code base against your test case and check the result, instead of running **YOUR** code against your test case. Please let us know if the leakage persists. thanks. PeiYong --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
