Yeah, I agree that they are more suited to designing new languages rather than parsing existing ones.
Parsing toy languages is not a big deal -- but then I have to make the observation that metalanguages are solving precisely the problem that's not a big deal.
But parsing real languages is a big deal. How many C, C++, Perl, Ruby, or shell parsers are there? How many text editors or lint tools contain poor implementations of such parsers?
Those are languages people actually use. I think the lesson is that humans can understand more complex languages than "standard algorithms" can, and they PREFER those languages (as shown by voting with their feet).
Metalanguages should be designed for the languages that people actually use rather than constraining them to an artificial model.
Parsing toy languages is not a big deal -- but then I have to make the observation that metalanguages are solving precisely the problem that's not a big deal.
But parsing real languages is a big deal. How many C, C++, Perl, Ruby, or shell parsers are there? How many text editors or lint tools contain poor implementations of such parsers?
Those are languages people actually use. I think the lesson is that humans can understand more complex languages than "standard algorithms" can, and they PREFER those languages (as shown by voting with their feet).
Metalanguages should be designed for the languages that people actually use rather than constraining them to an artificial model.