1.0 release of EYG IR and type safe eval.

EYG's Intermediate Representation (IR) is now stable. The spec for the IR format is available on github.

The IR is built on dag-json and so the hash references of expressions are also stable. This completes two key tasks on the roadmap.

Having a stable IR allows many things to be built on EYG. Want your own syntax? Build a parser to target the IR and use the rest of the EYG tooling to run it. Need a new runtime? Use the EYG editor and build your own interpreter or compiler. Fancy your own type system? Write your own and keep the editor and runtimes.

This flexibility separates EYG from most other languages where the surface syntax is the public API and IR/AST is an implementation detail that you can't rely on. A stable IR was always the goal of EYG as it saves tools from reimplementing parsing to build on the language. I can't wait to see what interesting things people build on it.

Safe code evaluation

Allowing arbitrary code evaluation would break type guarantees and be a security vulnerability for most languages. Implementing evaluation as an effect for EYG solves both problems.

In this quick video I walk through using the Eval effect and discuss it's implementation.

Here's the summary: An effect has explicit reference to a continuation that represents the rest of the program. By inferring the type of the code to be evaluated we can observe any side effects it might have before it is run. Inferring the type of the continuation can be used to check it is consistent with the evaluated value.

This effect is not part of the web runtime yet.

Other structural editors

This years Principles of Programming Languages (POPL 2025) had a few videos that caught my eye in their section on syntax and editing. In this case they are talking about structural editing - the same approach that the EYG editor uses.

The first was a talk about Pantograph an editor that can highlight expressions while leaving an inner expression unhighlighted. Their talk gives a much more visual explanation of why this is valuable.

The second talk covered Grove, an algebra for collaborative editing of AST's. This one dealt with theory but outlines a path to add collaborative editing in the EYG editor. It would also help with version control and the issue of merging concurrent changes.

It's great to see this work as it provides robust foundations for developing the EYG editor further.

That's all for this update, I hope you have enjoyed it.

This issue is available, and shareable, at https://eyg.run/news/editions/5