Mark - there are two issues: tspecials and the blank line.
In terms of the tspecials, at the next publication, we either need to
fix the example or fix the definition. For myself, I just can't see
what the big deal is about quoting it. Heck, my code quotes virtually
every parameter it generates. Is this really such a big issue that we
gotta change the tspecials definition? Why not just get people to fix
their code and also fix the example in the RFC?
I will argue that the blank line thing is less than optimal behavior,
an "annoyance", as you put it, but we are only talking about two
characters. As such, I really would prefer that we don't change it, the
BNF, etc., etc. Let's just learn to live with it.