For example, the devel/cppunit/.broken.html file contains a leading 
"<pre>", which is not closed. The file's contents is not passed through 
a to-html filter either.

Are there any objections to renaming the files to .broken.txt instead of 
.broken.html? And, of course removing the leading "<pre>"? The code that 
writes the "<pre>" is misplaced anyway, as it writes that tag already to 
the .broken file, which should really be considered text/plain.

Roland