Good link! Both article and comments hit on reasons JIT projects are so hard and fragile with CPython (e.g. wide scope/touches everything, hasn't gotten over the hump of being obviously better unlike JS's V8, the way CPython hands out raw pointers to C extensions, etc). Helps explain a bit why JS, Lua, Ruby have had more lasting success with JIT acceleration.

In this can-we-please-finally-get-past-the-GIL era, koflerdavid also makes a good point:

> CPython is currently stabilizing its new free-threading mode. The current budget for changes to the whole module ecosystem is therefore already strained. It's simply bad timing for other breaking changes to land since it blow up the test matrix too much.

Or the dynamic languages in the genesis of JIT research, Lisp, Smalltalk, Self.