Declan Naughton
Declan Naughton
This is necessary for planned implementation of renaming
For 90% models it should relatively easy to get from calculang bundles to modelx then to super fast executing Python and other places. Possibly "seemless" via Piodide. It gets hard...
But, I'm unlikely to complete the Python generation piece. Worth looking for existing JS-to-python tools first anyways.
mojo ([oss](https://www.modular.com/blog/the-next-big-step-in-mojo-open-source)) and rust are interesting targets; generator infra for a few targets could be pretty homogeneous
e.g. see inputs arrays here: https://calculang.dev/models/savings/savings.introspection.json
Bottup-up stats here give better perspective than screenshot above. The blob (brownish) is the actuarial calculations rerunning from 0 
Consider also: Ohm https://ohmjs.org/
Can do soon! (when I get around to do some environment setup)
nice possible application for llms, also useful mechanism to add structure to (otherwise structure-less) models
some reading: https://esdiscuss.org/topic/proposal-switch-expressions https://github.com/tc39/proposal-pattern-matching/issues/116 I thought there was a tc39 proposal for exactly this 🤔