http://www.mediawiki.org/wiki/Markup_spec should be a good starting point this.
Specifically one could adopt an approach similar to what the "Parser outline" section says and do some reverse literate programming (elisp--->Natual Language) Let me put on my didactic hat and remind the participants in this list/thread the "GTD approach" to Getting Things Done. Collect->Process->Organize->Review->Do I see (and I could be wrong) an intermingling of all of the above steps in a single step in this thread. A simple worg page that "collects" the low hanging fruits to begin with will be a good but firm first step in this direction. It would also enable consolidation of notes in a single place (rather than it being fragmented in a bunch of mboxy thingies) and also enable community contribution. Let's all step back from the mailing list (or this thread) for a while and contirbute to "growing" a spec from within a Worg page. I think it would help if we keep our focus away from lexers/parsers and bnf/ebnf etc and return to it when we refine the model. Formal Descriptions only adds precision and (possibly refinement) to a model. They are generally of limited help while growing the model itself. So a informal spec in plain English (with accompanying regexes) should be good enough. Mankind has made good progress with just pen and paper and lately with plain text files :-). ps: I definitely see Nicolas "at it". Others seem more like cheerleaders. Both have their roles to play and their respective contributions are equally helpful. Jambunathan K.
