Stream: cull-os

Topic: Why no IFD

RGBCube (Oct 12 2024 at 07:50):

Why no IFD is Good, Actually

• Faster evaluation
• No need to have a configured builder to evaluate the derivation graph
• Evaluator simplicity
• It will also enable me or someone else to create a LSP that interprets the expression with intermediate steps and gives you suggestions, without taking a year or doing anything impure or requiring a builder

Tim DeHerrera (Oct 12 2024 at 14:28):

Indeed, this is also the reason I am (eventually) planning to add a language agnostic plugin boundary to eka. I imagine most things you might want IFD for you can do more easily in the native language of concern, and then just feed them into the evaluation via a json stream or something

Last updated: Jan 30 2025 at 20:17 UTC