conspicuous example of newer tidy not handling errors as well

Hi,

I'm using tidy to screen-scrape the http://dmoz.org site (transformed to xml and parsed as xslt) and have recently moved from using the old COM wrapper (André Blavier's) to the new one. I've noticed that in at least one case the old version interprets a bad closing table tag more like a browser than does the new one. If you go to any listings page and view the source you'll notice that the unordered list of sites is not in a table. Also note that there is a bad table closing tag near the top of the page. The old tidy "correctly" closed the table, while the new one closes the table at the end of the body. I've noticed other cases where tidy waits until the end of the body to close a tag that could have been closed earlier.

Just thought this might be of some interest...

Lucas Fletcher
lucas@dealersinnotions.com
http://dealersinnotions.com