> my immediate reaction is that the pain of doing this is relatively
> trivial, compared to the advantages.

Ah, but I fear your opinion here doesn't count:-) I know you'd rather
have the authors validate their manuscripts against a DTD by SP or
some such, but the question is whether authors will do it...

If a `preprint' class for this markup *enforces* a suitably rich
markup by generating errors if insufficent information is provided
then it may not be too popular with authors.

If on the otherhand it takes a more permissive approach, for instance
not complaining too much as long as some kind of address is given in
either affiliation or address commands, and just using whichever is
available, then the useability of manuscripts prepared as preprints
with production classes that do *require* certain fields would be much
reduced.

I think that for a first approach I should aim to be `strict' but I'd
be interested to know what are the views of potential authors (as
opposed to publishers)

>  I would claim we have to distinguish:
> ..
Thanks, that is a useful classification.

David