Sometimes you can't avoid using "legacy" systems, and sometimes they are even good, but that is unrelated to the question of how a third party processor can be intregrated in a serializer.
Of course. And PDFs generated with TeX seem to provide much more quality than XSL FO processors at the moment, so it *is* an option to have a TeXSerializer.
If your original file is XML and you have an xslt that converts it into TEX text, the output from the xslt is passed as a series of character sax events, so even though it is technically xml/sax events, since there are no elements or attributes, you can easily treat it as a text stream which can be fed by a serializer to an external processor if desired.
NO! Exactly this is not an option. We won't break the sitemap contract for having a specialized serializer. After every step in the sitemap it must be possible to get well-formed XML - and that's not possible if you have text as output of a transformer and as input for a serializer.
As Bruno and me pointed out the TeXSerializer must be able to read an XML representation of TeX. Only in the prefered solution he has another opinion than me.
Joerg
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]