Working on Rewriting the Preoptimizer Preopt.ohm

2026-03-29

Mar 30, 2026

I don’t want to deal with Substack’s inferior formatting capabilities for program code, so I’ve converted this article into PDF form and am providing it as a downloadable file.

TL;DR

This article is about my ongoing work on the Frish project - namely fixing the pre-optimizer stage. At this point, I need to throw the existing code away and start again. This article contains a justification for discarding the code, and, how I make it “easy” for me to use this option. The point of designing a program is to discover the details of the problem space in an iterative manner, not to create production-ready code (yet). I discuss my thought process and the tools that I use to help me iterate and toss code away. I only discard code, but retain what I’ve learned. The goal is to use tools that make it easy for me to spend more time thinking and learning rather than polishing code - all the time knowing that I can throw all of the code away, since it hasn’t cost me much effort to create. This approach is different than that of using a strongly typed programming language. First, I want to figure out what I want the code to do, then I can re-cast it into stone using tools and compilers meant for Production Engineering.

L;R

…

What I’ve learned, thus far

…

Refreshed Approach

…

Repository

The WIP code is on github https://github.com/guitarvydas/frish/tree/debugdefsynonym.

An accompanying twitch-like (long) video.

I will write new code. I find it difficult to write prose about what I’m thinking or to talk into a camera while I think. If you want to follow along, see the code in the repository and/or visit the “programming simplicity” discord and poke questions at me.

PDF