On Tue, 06 Jun 2006 21:52:37 +0100, Doug Bryant <[EMAIL PROTECTED]> wrote:
> Ross, > > I ran across xml-smarty today. It is also based on libxml and the author > seems to have solved the cross document node copying issue. > > xml-smarty's license is lgpl rather than the MIT license, but I believe > it > is legal to look and see how things are done as long as you don't > directly > copy the code. Then again, I am no lawyer :) > > http://raa.ruby-lang.org/project/ruby-xml-smart/ > > I hope this is of some help. > Yeah, I've looked into XML-Smart on more than one occasion to see how things are done ;) On this occasion, though, there's not a great deal of difference in the way we're doing things - I think this problem is more rooted in the mixture of data pointers and pointers to data pointers floating around inside the Ruby XML::Node instances, which comes from, well, pretty much everywhere :) I wrote up a non-ruby test for the xmlDocCopyNode idea I previously suggested, and found that without copying the node first I saw the same problem as with the extension, but by copying first the problem went away, and valgrind gave a clean leak-check. XML-Smart makes do with xmlCopyNode to handle this, but still I feel sure we need to look deeper at the way we're handling those pointer nodes. -- Ross Bamford - [EMAIL PROTECTED] _______________________________________________ libxml-devel mailing list libxml-devel@rubyforge.org http://rubyforge.org/mailman/listinfo/libxml-devel
