Today was one of those fascinating days when you spend eight hours hunting down stale cache and redundant model associations in huge and quite legacy codebase.

This weekend I finally finished working on Heroku buildpack for / Joy framework. The background job using jurl (curl wrapper for Janet) works with Heroku scheduler. Still need to write some readme and properly cache curl upgrade compilation.

So I built my first "web app" in

I'm not gonna lie, this wasn't easy, but in the end quite rewarding. Lot's of low-level stuff (for example pg driver does not handle date, because Janet does not have date type, so you have to do it manually).

It's been a while since I wrote a blog post. Here's one from few days ago about using profiling tools to understand how Ruby apps work:

Interesting thread on Twitter about Rails, Spring and Bootsnap:

Personally I think Bootsnap is pretty good, although it does not need to be added right away for every Rails app. Spring on the other hand only causes problems.

Current status: trying to write PEG grammar to parse SDLang documents using Parslet.

Links for reference:


This is surprising, but Zoom works sooo much better on Linux than on MacOsX.

I decided to create an account on, so: hello all!

I had previous mastodon account but nothing really going on there, so I probably won't bother with a proper migration.

A Mastodon instance for Rubyists & friends