On Sat, Sep 6, 2014 at 2:21 PM, Jonathan S. Shapiro <[email protected]> wrote:
> On Fri, Sep 5, 2014 at 6:31 PM, Ben Kloosterman <[email protected]> wrote:
>> "ad hoc parse rules and/or parse ambiguities." are common in nearly all
>> languages - with good reason they have taken short cuts to get something out
>> the door.
>
> For many languages that is so. None of those languages are languages that
> anyone has attempted to do regular formal reasoning about successfully.

I agree with your points, but have you considered making the AST the
normative format for BitC programs, for the sake of formal reasoning?

Not that it's impossible or even impractical to formalize surface
syntax, but for reasoning about surface syntax, the first step would
probably be to apply a verified parser anyway. If the AST is
normative, you get to skip that, and your spec is reusable with
different surface syntaxes.
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to