The Translator's Guide does not mention the use of HTML "entity references", i.e. escape sequences like:
à
ö
Apparently, it was assumed that it was not necessary to enforce the use of these entities, and to rely on the conversions done by the browser...
I see that the xml files in the catalog directories contain the UTF-8 encodings of these "extended" characters... On the other hand, characters like '<' are encoded with: '<'. I have found an exception to this rule: in the italian translation the entities are used, but in the wrong way and a word like 'Comunità ' is in fact encoded as 'Comunit&agrave;'.
I think there are 2 options:
1) keep UTF-8 in the xml files (this seems to work so far but I suspect there may be problems dependent on the browser),
2) allow entities in the xml files (but I believe the translators do not have access to these files).
Giancarlo
