[bitc-dev] Random thoughts on mixfix

[Aleksi Nurmi]

2010/3/18 Jonathan S. Shapiro <shap at eros-os.org>:
> where the semantic actions of mixsequence simply build a concatenated list
> of the constituent mixfix keywords, identifiers, and literals. We then run
> ResolveMixFixSequence on this list as a post-pass, which converts the
> initially parsed sequential AST into the expression AST intended by the
> prevailing mixfix syntax rules.

This sounds a lot like what is described in "Parsing Mixfix Operators"
by Danielsson and Norell as a possible implementation approach (they
also mention it's already used in Haskell). I think it's safe to say
that the approach does work, but as they point out, the parser must be
able to find the extent of the expression somehow.

> Anybody feel like taking the time to play around with one approach or the other to help move BitC forward?

What would you like to know about the first approach? I know nothing
about GLR parsing, so I think I'll pass that one.