I'd like to close ISSUE-10 today so we can make progress.
My strong suspicion is that the pattern proposed earlier*
based upon names in Python etc is going to provide the
most interoperability with current databinding implementations,
our criteria for a pattern to be "Basic".
However, I recognise that we should go the extra mile to
avoid disallowing i18n characters, especially since we are
working within the W3C!
I therefore propose we have at least two patterns, one to recognise
simple names* for name="" and anothers to recognise more
"interesting" names (unusual in terms of mainstream programming languages),