Paul Ganssle @pganssle@qoto.org

Anyone have a good example of setting up GitHub Actions for old, out-of-support versions of Python (e.g. 2.7, 3.6)?

If anyone else finds this kind of thing useful, I'd totally love it if someone else started using this project. Particularly if you are the kind of person who is going to make lots of improvements to the front-end and then send me PRs 😉

I keep coming up with interesting improvements for this project, but I only have so much time to work on stuff like this.

I started this application in December 2016, before I knew anything about databases, so I hacked together a pseudo-DB out of YAML files, because I wanted to be able to edit the files by hand if I screwed up. As this "database" grew, parsing huge YAML files became a bottleneck; I lived with this for years, but recently, I managed to switch over to using a SQLite database!

I lived with this for years, but recently, I managed to switch to a SQLite database!

This was surprisingly easy, because I already had a pseudo-ORM, and I just load the whole "database" into memory at startup, but I am still not using the features of a "real database", since my "queries" are basically Python code iterating over dictionaries and such.

I really like the "segmented" feed, which breaks up books along chapter and/or file boundaries, recombining them to minimize total deviation from 60m files. I like to listen to audiobooks in ~60 minute chunks, and this automates the process of chunking them up for me.

The implementation was a rare example where dynamic programming was useful in the wild (and not just in job interviews): https://github.com/pganssle/audio-feeder/blob/1a07c8ffa7c7b548471f979382fedb653ce6ee5a/src/audio_feeder/segmenter.py#L45-L102

Thanks to @njs for suggesting the approach and basically implementing it flawlessly on the first try.

I've also created this probably convenient docker-compose repository for (somewhat) easily deploying `audio-feeder`: https://github.com/pganssle/audio_feeder_docker

Now featuring ✨🌟✨*installation instructions*✨🌟✨ (so fancy).

Yesterday I released version 0.6.0 of my audiobook RSS server, `audio-feeder`: https://github.com/pganssle/audio-feeder

It takes your directory of audiobooks and generates an RSS feed for each one, so that you can listen to them in your standard podcast listening flow.

I'm particularly happy with the new feature "rendered feeds", which uses `ffmpeg` behind the scenes to generate alternate feeds where the audiobook is broken up along different lines.

#projects #python #ffmpeg

I was thinking about this because I was thinking about how I might optimize the resource consumption of a program where I'm ~the only user, and then I thought, "Hmmm.. If I spend that same time working on an open source project currently used by millions of people, with even a modest improvement I could probably save more energy than my homebrew project will consume in its entire lifetime."

Fun stuff.

It would be interesting to live in a world where most people used this or something like it in production: https://www.sciagraph.com/

I'm kinda curious to know stuff like, "How much electricity would it save if time zone conversions in pandas were 20% more efficient?"

@hynek FYI, `structlog` seems to be affected by this.

Hm... PEP 621 deliberately doesn't have a replacement for `setuptools`'s `license`, but they do have a `license` key that is equivalent to `license_files`. Unfortunately, this is what PyPI does with the data in that field:

Not great.

Discussing the latest Jeopardy happenings with my family over Thanksgiving is the closest I've ever felt to knowing what it's like to be a sports fan.