(No comment was entered for this change.)

I have incorporated your algorithm into jGrouseDoc (see it in svn) with minor
modifications.

So far I have found one problem though - it does not handle properly valid JS
identifiers like $break or _continue and the like since the first pass removes all JS
reserved words.
Also it would try to parse strings, so construct like
/**
* var bla
*/
"SOME STRING".indexOf(...) 

would result in parsed id SOME (not that one should expect this kind of code
somewhere, but nevertheless...)

Can you take a look how this could be addressed (since you started that :-))?

BTW, after adding the check to jGrouseDoc I have cleared ~20 doc errors in my
framework :-)

okay, I tweaked the reserved word regex to handle things like $break and _continue (I
hope).  I haven't actually tested it, but I'm pretty sure it'll work.

Re the string literals ... hmm... that's the sort of thing I think we can punt on for
the time being.  I think the current parseIdentifier() behavior will deal with 90+%
of the most common cases, which is a huge time savings.  I'd rather wait until we
have concrete examples of other (common) code patterns that need to be addressed
before trying to tune it's behavior.

(No comment was entered for this change.)