But speaking of HaXml bugs, I'm pretty sure HaXml doesn't handle % correctly. It seem to treat % specially everywhere, but I think it is only special inside DTDs. I have many XML files produced by other tools that the HaXml parser fails to process because of this.
I had a similar problem where the parser choked on % signs in attribute values in the XML file. I solved it by playing around in Lex.hs. I could check for the (ugly) solution...
Yes, it is expected. An ampersand indicates the start of a reference, e.g. < or If there is no semicolon to indicate the end of the reference, then it is a parse error. The XML specification is quite clear that neither & nor < are valid standalone characters in an attribute value.
Which is exactly the problem. The & is part of a reference, namely ë. Regards, Koen.