Published on 7 September 2021.
This is what I’ve been up to in August 2021:
I continued with the draft post RLMeta poster 2.
I had some ideas of things I wanted to improve in the next version of RLMeta. But mainly, I just wanted to work on it again because I think it is a fun piece of software to work on.
Before I could continue to improve RLMeta, I first had to port it to Python 3 because my new laptop now runs Python 3. The port was rather small.
I then continued with various improvements that I plan to document in the blog post.
I find it difficult to know when to stop improving RLMeta. Sometimes I feel like I chase perfection. In every iteration, I get closer, but there seems to always be something left that annoys me.
My next big goal with RLMeta is to make it not depend on Python. In order for that to happen, I need to implement a subset of the Python language that RLMeta uses. All the way down to machine code. So it is a quite big project, but should be fun.
I thought about where it makes sense to put CI/CD code. The most common place to put it in (that I’ve seen) is in the repository itself. But I feel like that is not the most logical place. Information about how to use a project in a CI/CD chain should not be coupled with the project itself. I feel like build scripts belong in the repository because how to build a piece of software feels related to the project itself. But in CI/CD, you often deal with larger aspects than the project in isolation. For example, in a CI/CD chain you might want to send email notifications upon build failure. That is not related to the project. It is related to how you implement CI/CD. So I’m thinking that a CI/CD system should be programmed separately from the project repos.
I got inspired by this video by Van Neistat about a tool he can’t live well without: the pencil. I could relate to that as I also feel that the pencil is “the pump that brings up the ideas from the well”, so I decided to buy a T-shirt:
Site proudly generated by Hakyll.