> Kevin's argument that a context free grammare(*) is our best tool to express
> what the difference between an LHS and an RHS is seems very weak.
> The restrictions on an LHS properly belong to the world of *abstract* syntax.
My point was really that a formal mechanism for expressing concrete
syntax is the best means of ensuring implementation consistency at the
character level. This is unfortunately still important... And a CFG
is the best tool I know for this purpose.
The introduction of n+k patterns complicates matters by overloading the
concrete syntax of LHSes. It is thus necessary to disambiguate this
syntax in order to distinguish the denotations of the abstract forms.
Surely this can't be done in abstract syntax?
Kevin
