100% agree on the current instructions not focusing in optimized execution.
The note about placing the main loops higher in the code probably could easily be done. One of the things I was focusing on, was just getting code that was error-free, not necessarily optimized. In my mind, I’d use the LLM to generate the working code, and then use the tooling for minor optimizations. Right now, I can basically replicate the Minification of Cleuseu, to trim down on wasted bytes. I could probably add a feature that does a find and replace of all variables in a program, with single-letter vars, no sigil to save space. -George On Wed, Dec 24, 2025 at 12:42 PM Joshua O'Keefe <[email protected]> wrote: > On Dec 24, 2025, at 4:33 AM, [email protected] wrote: > > https://github.com/Grimakis/m100-programming-skill > > > Hi, George. I'm planning to work with locally-operated models so I'll be > using a trimmed-down version of this; most front ends don't have the > ability to do conditional includes, even SillyTavern. Having said that, I > spotted a couple spots where you might want to make changes depending on > the output you're looking for, particularly if a program is primarily BASIC > with limited to no ML. > > > Prefer explicit typing > Note that implicitly-typed variables when used as integers are faster to > access than typed integers, at least on the 100/102. I've never looked into > why this is. Nor am I clear if this is the case for floats. I'm not sure > what you're using as a linter, if anything, so that could be a > consideration if performance is a factor. > > > Section spacing > Another optimization that's easiest to build in early is to place > subroutines earlier in the listing. GOSUB searches linearly from the top of > the listing for the destination so calls are faster to lower-numbered > lines. It's possible your tooling could do line renumbering without the > pain of on-system tools which might make the optimization pass easier to > defer. > > Thanks for putting this together. I'm looking forward to using it as the > basis for a less complex self-contained system. >
