I'm curious - how did you test it for compatibility with the main K implementation?
I'm curious - how did you test it for compatibility with the main K implementation?
I don't have a rigorous approach for testing compatibility at present; My starting point was the K2 manual, and I've tried to get as many examples from that working as possible. I have been learning the language as I go, so often I don't realize functionality is missing until I stumble across an existing K program that doesn't work properly.
I'm targeting K5, which is unreleased, so there's a great deal of speculation. I think I have a decent subset of the functionality working now, but most of the time when I write new programs using it I still shake out bugs and unhandled edge cases.
Since you are playing in JS, you might also have a look at c.js (http://kx.com/q/c/c.js), which is the tool KX uses to translate between JS objects and k objects through web socket IPC with a k process (http://code.kx.com/wiki/Cookbook/Websocket). Might be useful to see how Arthur envisioned the k data structure in JS. Also, it would be kind of interesting to have a k5 process running in the browser talking with a k4 process running locally in the 32bit trial...