I don't think clpfd or chr are especially large hogs of memory. Have you run into issues there? If you can write a smallish program that uses a subset of J (ac

I don't think clpfd or chr are especially large hogs of memory. Have you run into issues there? If you can write a smallish program that uses a subset of J (actually APL might be easier, since it is less likely to interfere with SWI's parser) I would be willing to try and figure out how to run it. Aim for 5-10 lines. Email me. I like the idea I'm just not totally convinced. I feel like if I got better at J, my problems with it probably would be resolved. For instance, it feels to me like J is really bad at parsing (beyond regular languages anyway) and tree structures. But there are built-in verbs for these things, so I could be mistaken. I only threw a few months at it last year. And I kept being indecisive about APL and J. It didn't help that Aaron Hsu was such a passionate advocate for APL, and Roger Hui apparently now works for Dyalog and has tried to close the loop on a lot of the ways J improved on APL there. My back-of-the-envelope sense, from writing a small amount of J and a fairly significant amount of Prolog, is that J is way better at solving the core of a certain class of problem, but you spend a lot more of your life converting the input into an internal representation you can use and back to a result a human might want to see. Prolog really likes a certain class of problem which is somewhat less general than J, but it makes up for that limitation somewhat by being amazing at parsing and having great internal representations of things. In most of the cases where I would want non-determinism in J, I would probably try to rework my problem so that I'm doing all the possibilities at once and then detecting which ones succeeded. If I had non-trivial calculations to do, I probably wouldn't use Prolog, but if I had to, I would certainly consider calling out to C or something. My day-to-day interests seem to coincide more with what Prolog does well than what J and APL do well. Anyway, I think it might be a profitable area to explore. I was curious about Brachylog for similar reasons. https://github.com/JCumin/Brachylog https://github.com/JCumin/Brachylog Anyway, I'm curious if you could show me what you are thinking. There must have been a smallish program you thought would work well with this combination. Can you share it?