I'm posting an update on this PR. While the migration to Expronicon offers potential benefits like recursive algebraic data types and wrapping numerical values in BasicSymbolic, the work involved has proven more engineering intensive than initially anticipated. It has resulted in numerous broken tests, both within SymbolicUtils and in downstream packages, and debugging these requires a significant time investment per test.

At this point, I don't have a clear timeline for completing this migration, and the performance implications are still uncertain. Due to other priorities, I'm currently putting this work on hold.

I'm happy to hand this off to anyone interested in pursuing it. The PR contains the current state of the work, and I'm available to answer questions and provide context. Please comment here if you'd like to take over or discuss further. Any help would be greatly appreciated!

Hi @bowenszhu! I originally wasn't intending to spend any time on this, I just came across it out of my questions in JuliaSymbolics/Symbolics.jl#1354 along with my interest in implementing non-commutative symbolics for applications quantum info. However, I had a some free time this past weekend to take at look at this and also all the various approaches to implementing ADT in julia. So I thought a good exercise for myself to better understand how the SymbolicUtils works would be to try to do the same thing this PR does, except with Moshi.jl. The result of my hacking on this this last weekend is #676.

Anyways, I don't necessarily have the bandwidth to move forward on this very quickly. But if it does seem like something that may turn out to benefit julia symbolics and get merged I do have interest in continuing to pursue this.