qemantra⌗

I have finally released v0.2.1 of my project qemantra. This release removed a lot of code, made it cleaner.

Things I have learned while developing this release.

• How to build a sexy website for any project. Using vhs and mdbook, I made a website inspired from WezTerm, mostly it’s homepage. I tried to include a visual element (Picture or GIF) to improve user engagement and understanding.

• How not to complicate argument parsing. When using flaggy, I used to have a separate struct for each subcommand, over the structs I had for actually running the program. With this, I eliminated flaggy and brought the good’ol flag package. I also tried to eliminate almost all other structs. Now I have a single struct for the entire project. This simplifies things a lot.

Things I could have improved

• Testing: Even though I don’t write a single line of code without testing, refactoring made lot of the tests not-passing. Causing my CI/CD (Github Actions) to fail on every commit. I would have to re-write these tests.

• Website deployment: Currently deploying the qemantra website is a chore. I need to build, archive, transfer to my server (Using sftp) and then unarchive and restart my reverse proxy. This is a lot of work and I would have to improve the workflow by a large margin. This is my plan over the next few days.

• Better planning: This release was a little mess, the website had a lot of references to the v0.0.1 release, I had to do a little cleanup for the few days after the release. Next time I can have a ready changelog and beautiful CI/CD pipelines to make the process easier.

Try it here

bsharp or b#⌗

This is my Lisp implementation inspired from bash. I finally built a hand-written recursive descent parser. My last implementation boombash had no actual parser. It had a very rude implementation and had no actual AST creation.

With the current implementation I have finally created a proper AST. I have also started the evaluator work. It can evaluate strings, numbers and other simple objects.

Ideas I have for this project

• Write a compiler: Once I have a working interpreter, I can have convert it into a working compiler. I am learning this while writing a VM based compiler uranus ( My implementation for monkey from Writing a Interpreter/Compiler in Go.).

• Writing a transpiler: As b# is inspired from Bash and aims to replace bash scripts, transpiling to Bash script sounds like a good idea! With this evaluation shifts to the much better bash implementation.

• Adding a macro system Writing a Interpreter in Go has a secret chapter called Writing a Macro System. In keeping with this, I am also interested in making a macro system for b#. Anyway every lisp is incomplete without a macro system!

Try it here

You can also checkout uranus and boombash

My goals for next few days/week.

• Work on publishing groom with a sexy website.
• Complete groom-install, bringing number of subcommands to 3.
• Maybe implement task dependency ordering using a topological sort for groom-make.
• Finally complete dpm and bbx.