Trinary logic on top of "ordinary" Lisp logic?

[amno1]

Hi,

I saw postings about Elsa on the mailing list, and checked it out yesterday.

I think it looks like a great idea and very interesting project. Mostly because, once you have a static analysis, you can do more than just report errors: rewriting (compiling) lisp forms is not far away. For example a Lisp compiler should detect that mapping functions are just for-loops and generate code that eliminates (inlines) both function applied to each element and mapping function itself.

Anyway, this trinary library itself was a little interesting. I thought it looked a bit unusually C-esque, or perhaps in this case JavaScript-esque. I have just rewrote it without using descreete logic (-1, 0, -1) by using generalized booleans from Lisp. I don't know if that works for Elsa, wasn't able to test due to Eask refusing to find tests/elsa-test-helpers.el, but as soon as I get that to work I'll try it in the Elsa context.

Meanwhile, all tests in trinary-logic pass (using Eask and Buttercup). I don't know if that is of interest to you, but I believe it should be slightly more efficient computations.

I have also somewhat "relaxed" "int to trinary" to work more as a generalized boolean:

(cond
   ((> x 0) (trinary-true))
   ((= x 0) (trinary-maybe))
   ((< x 0) (trinary-false)))

That works on any numbers, not just integers, but can easily be converted to work only on fixnums, and with -1, 0, 1. In the context of Elsa, I believe you could have done without trinary-logic at all, since Lisp(s) already have, sort of, trinary logic built-in, but that would be a different discussion, and of not so much practical interest.

Just an idea and a suggestion. If it is of interest I can give you a PR.