In my experience, HTML files DO currently switch -- to
Under those circumstances something went wrong with http://www.gutenberg.org/dirs/1/4/3/4/14340/14340-h/14340-h.htm .
However, the text files use -- because the entity equivalent doesn't exist in 7bit ASCII.
That's okay.
I think I've seen this discussion before on DP forums. If I remember correctly, it was decided to stick to the xyz--xyz standard simply to avoid confusion and complication.
I'm not sure whether the German reading community will get used to it ;-)
Just to avoid to avoid complications ;-) a) \227 for em-dash is neither ascii nor iso-latin nor unicode: it is windows codepage. em-dash is — b) unicode has two slightly different characters, em-dash and horizontal bar, ― the latter is explicitly indicated in dialogues--that is where it is mostly used in german, french and italian, and other languages. It would make a lot of sense to use horizontal bar (that has some space around) and em-dash (without) where they are indicated. To forget en-dash – figure-dash ‒ etc. But, as I said, this a further level of complications (i.e. typographical precision) that is probably beyond our (present) reach. Carlo Traverso