This is a neat project which makes the λ-calculus more accessible. The λ-calculus is incredibly useful for designing and analyzing language features, but it can be a bit of a pain to work with in part because it quickly becomes verbose. Papers and proofs usually end up developing their own shorthand, but that tends to be rather <i>ad-hoc</i> and is rarely implemented programmatically.<p>In my view, this makes this project a great platform for playing around with language design. It should be reasonably easy to take the existing base and extend it with additional features (exceptions, continuations) or types. I've long suggested writing a handful of λ-calculus interpreters, from untyped to simply typed to System F to dependent types as a great way to thoroughly learn programming language concepts; Caramel gives you a great starting place for this coupled with an accessible, readable syntax.<p>Also, it would be great to see an online version of this, perhaps compiled with GHCJS. I've played around with trylambda[1] which is fun, but the lack of syntax quickly gets in the way. Caramel would be a great alternative.<p>[1]: <a href="https://trylambda.com/" rel="nofollow">https://trylambda.com/</a>
Would love to see this compile to blc
<a href="https://en.wikipedia.org/wiki/Binary_lambda_calculus" rel="nofollow">https://en.wikipedia.org/wiki/Binary_lambda_calculus</a>
I don't think I understand the reasoning behind the ADT. Isn't a 'type' just a set of predicates applied to arguments at compile time and/or at runtime? Why not just define a Caramel convention where you specify typed functions using a fancy argument list, in this case, a list of lists of predicate functions? Untyped functions would take a different shape of argument.
They use .mel extension, which reminded me of the old <a href="https://en.wikipedia.org/wiki/Maya_Embedded_Language" rel="nofollow">https://en.wikipedia.org/wiki/Maya_Embedded_Language</a>
Had never heard of the lambda calculus before. Very cool.<p>Does it have any practical applications? Or is it mostly a theoretical / conceptual tool?