> i have to build a collection from plain text files which contain
> non-ascii characters - originally they are encoded in ISO-8859-1
> (windows ansi).
>
> the problem now is that i use these files to create a metadata.xml
> by extracting text and inserting it into meta tags. as a consequence
> this yields a "not well formed" metadata.xml!

Hi,
in Greenstone the metadata.xml files must be encoded using UTF-8.

Hope this helps
John McPherson