> I looked at the sources this morning and failed to realize the
> documentation was an simplified explanation of the actual implementation.
That's where my reply should have started. It's been awhile since I've studied those regular expressions and I messed up by not consulting them before I tried to explain the matching rules in my reply. But regardless of how I explain it, it's the pattern that we ultimately have to reconcile with. I agree that the spec is a good opportunity to clarify what the rules are, and perhaps even simplify them if feasible.
> If I understand it well, constrained formatting marks are allowed at
> "word boundaries" where a "word boundary" is a defined as the transition
> from a non-word to a non-space character (or the other way around for
> closing markups).
That's generally the idea, but still not an exact explanation. That's because it's possible to use constrained formatting around a sequence of characters that are not words at all. For example:
*!?$@*
Another way to explain it is that there has to be a space or start of line on the left and a non-space on the right of the first formatting mark (and the opposite for its pair). But that rule may prove too loose to specify and thus may need to be challenged. I don't know enough about the implications to take a position on that yet.
Best Regards,
-Dan
--
Dan Allen, Vice President | OpenDevise Inc.
Pronouns: he, him, his
