I’m not sure if it will be of an interest for Crystal directly but it looks like quite unique JIT compiler related to making Ruby 3 fast.
Author mentioned he will validate idea and implementation with Ruby first, but it’s designed to work with other languages too. He also mentioned interoperability with languages like Crystal via LLVM. The WASM target should also be easy to achieve.
Nice. I’d always wished there were some kind of speeder-upper for Ruby…I guess like python’s pypy or whatever the older equivalent was called. Hope it speeds things up for them.
Reminds me of this project: https://github.com/rdp/crystalizer that worked on ruby 1.8 10 years ago LOL. It calls out to gcc though…
If they can make it work they’ll finally have solved the ruby puzzle…rubinius was also supposed to do that…and that other smalltalk version…so I hope they do finally…must be non trivial…
Yeah I hope so too. Was tracking and hoping for things like Rubinius too. Fast Ruby is long overdue :)
It seems for the approach taken in MJIT, where they run the C compiler instead of using an API, the tiny startup cost of MIR would be crucial. Rails with MJIT might actually become faster.