Light summer post, want more then follow the links.
1. after a chemlambda demo appeared on hackernews (link) I saw a lot of interest from hacker brains (hopefully). Even if slightly misplaced (title reads: D3.js visualization of a lambda calculus) it is an entry gate into this fascinating subject.
2. Sreejith S (aka 4lhc) works on a python port for chemlambda, called chemlambda-py. It works already, I look forward to try it when back home. Thank you Sreejith! Follow his effort and, why not, contribute?
4. During discussions with Sreejith, I noticed some funny facts about the way chemlambda computes the Ackermann function. Some preliminaries: without cheats (i.e. closed forms) or without memoization, caching, etc, it is hopeless to try to compute Ack(4,2). The problem is not as much the fact that the function takes huge values, but the fact that in order to compute even modest values, there are lots and lots of calls. See the rosettacode entry for the Ackermann function about that. Compared with those examples, the implementation of chemlambda in awk does not behave bad at all. There are now several mol files for various ackermann function values which you may try. The only one which takes lots of time (but not huge memory, if you except the html output, which you can eliminate by commenting with a # all printf lines in the awk script) is ackermann_4_1. `This one works, but I still have to see how much time it takes. The interesting observation is that the number of steps (in the deterministic version) of chemlambda (mind: steps not rewrites!) is at 1 or 2 difference than the number of steps of this beautiful stacks based script: ackermann script. It means that somehow the chemlambda version records in space the intermediary values, instead of stacking them for further use. Very strange, to explore!
5. There exist, on paper, the chemlambda v3 “enzo”. It’s very very nice, you’ll see!