From: Daniel Veillard
Well if you have a new patch which works with CVS, please send it,
though
creating that new xmlReaderFreeID() is the right solution.
Attatched is the new patch. I was able to run the regression tests on a
laptop and they appear to be fine.
the changes include the following:
xmlFreeProp
- removed the checking for intSubset, extSubset and doc
the doc is checked in xmlRemoveID and the others arent required to
add an id
- check is now just cur->parent != NULL and xmlIsID(..)
xmlUnlinkNode
- in the attribute block, xmlRemoveID was added as IDs need to be
removed if an attribute is unlinked
added xmlReaderFreeID function
added a static removeID function which both xmlRemoveID and xmlReaderFreeID
call since the code is the same other than the line on how to deal with the
xmlIDPtr
xmlTextReaderFreeProp
- uses the xmlReaderFreeID function
- did not change the test for intSubset or extSubset as it didnt seem
that IDs could be added on the fly like in DOM.
Let me know if anyone sees any issues with this patch.
Thanks,
Rob
Attachment:
libxml.diff.txt
Description: Text document