This is not real bug but - at least I hope - a sensible extendsion of the HTML parsing.

In order to support the HTML 4.0 successor XHTML1.0/1.1 the parse would be required to interprete empt< tags. e.g "<code />" should be interpreted as "<code></code>" and not as "<code>".

code.html:
#
# <code />
# static assert(0);
#

I agree this is a border case but some dump html editors produce this kind of output after deleting the original content between the
code tags.

test case: svn://svn.kuehne.cn/dstress/compile/xhtml_tag_01.html

Thomas