Taylor Monthly - April 2026
| Count | |
|---|---|
| Releases | 0 |
| Commits | 9 |
| Contributors | 1 |
| Pull requests merged | 0 |
| Issues closed | 0 |
At a Glance
- Taylor's Five Year Anniversary It's been five whole years since I started Taylor
- Refactored Taylor's Config handling Noticed this area was really bad and cleaned it up
- Taylor Export Woes Found an issue with my exports
- Taylor Squasher Brainstorming around implemntation
- Taylor Has a Community Site It finally happened!
Taylor’s Five Year Anniversary
I wrote a whole blog post so go check it out!
I’m just so blown away at how far I’ve come with Taylor. It even lead to me getting my current job!
Refactored Taylor’s Config handling
An issue was raised because I have a discrepancy between the
documentation and the code. In the code taylor-config.json is the file used
for configuring all your Taylor needs. In the documentation I call it
taylor_config.json. An easy fix right? Just rename the documentation? Well
yes, that did fix it.
I looked closer at the code though because I briefly considered supporting both as a politeness. I did not end up doing that but I did notice how awful the configuration handling is inside Taylor. So I took some time and completely refactored the configuration handling and created the Taylor::Config class to help out. This has reduced the amount of magic strings across the codebase and now lets developers easily use this configuration for themselves if they wanted.
Taylor Export Woes
The build for my previous config change broke on the web test suite. I thought this was really strange and was able to reproduce it locally. It just boots up, runs the C++ code portion of Taylor, then does nothing, and exits.
This took a decent chunk of debugging, lots of puts statements, and even going
back commits. But the strangest thing was the mrb state didn’t claim to have
an exception when running the mrb_load_irep method to load the test suite.
Interestingly going back to earlier commits didn’t fix this and it started
happening on my Linux builds too. This made me realise the actual cause. It was
because I was using the master branch of MRuby in the export Docker image but
I was locking it to the 3.4.0 tag when I compile libmruby.a. So the most
recent commits to MRuby seem to have changed how the bytecode is generated and
loaded. This meant the older 3.4.0 version of MRuby couldn’t load and run it.
The fix isn’t that interesting, it’s mostly just me locking the MRuby version so this doesn’t happen again.
Taylor Squasher
I’ve decided to pick this project back up and give it a serious attempt. I do believe there’s some huge value to be had here. Right now I’ve mostly just sat down and thought about what I want to tackle and how I want to tackle it.
I think I have some decent ideas that’ll let me implement require_relative
too, but that’ll take a little experimenting before I’m confident.
Taylor Has a Community Site
You read that right, Taylor officially has a community site! I did the big announcement over on this blog post if you want the details, or you can just go check it out!