Support for the XML tag has not been fully implemented yet. I remember you two (rickg,kipp) talking about this once, but I forgot the details. The URL given contains a testcase (at the very bottom).

Currently the parser kicks the markup out of the XMP tag as if it couldn't contain when instead it should ignore markup until it sees either the end of document or a </XMP> (I forget which - a simple test with navigator will answer that question). We should probably treat it like TEXTAREA and SCRIPT and so on and use the skipped content *OR* the parsers should treat it like a container and then add everything inside it as text (I like that answer better). The same goes for PLAINTEXT, except that the EOF conditions are different than XMP.

XMP support is now in, along with support for plaintext. They're not perfect, so other bugs may surface.

Fixed in Build Release 11/30.

regressed?

works again. must have been a manifestation of bug 40408.