RSS

Recommendations on ramping up on large software projects while managing complexity and showing results, learned the hard way.

Introducing my slip-box: https://explog.in/slipbox/.

I take org notes in logseq; which are then processed with a tiny python script & published using Emacs's org-export with an automatic Github Action every few hours.

After spending a large part of my career building tools, I realized I had unconsciously followed a consistent framework for identifying and building developer tools. It seemed like a good idea to make it concrete – both for sharing and collecting feedback – leading to this series of posts.

I've been tweaking this series for a few months at this point, and it's really exciting to finally call it done (for now).

Seeing RSS entries published in Recurse Center's aggregator made me realize that the links/entries being published in the feed were broken. This is an attempt at fixing them.

I've reworked how expLog is generated to make it easy to publish by a Github Action, with a shell script to run emacs in batch mode to make this significantly easier.

The RSS feed became a casualty of this work, but I finally got around to fixing it today, and I hope it gets picked up seamlessly for anyone who had subscribed in the past.

PS. I've been publicly iterating on a series of posts on building developer tools; I'm still editing, but I should be finished soon. Let me know if you have any feedback!

A quick introduction to Advent of Code for those of you who've never tried it out; as well as vignettes of what I learned from participating in AoC 2020.

Belatedly realizing that it's surprisingly simple to implement a neural network from scratch, particularly if you don't care about using it in production – some developer notes and a snapshot of a basic net in roughly ~150 lines of Rust with minimal dependencies.

(Unrelated aside: A Notebook Style Guide (Slides) – a short talk I gave at JupyterCon 2020 was released on YouTube. It started as a blog post here earlier this year.)

Panopticon is a Python tracer I've written as an alternative to using a debugger. Instead of manually stepping through code, I'd much rather have a computer do it for me.

Since writing this post panopticon has evolved to include probes: which can instrument code you might not have full control over.

(I'd initially written this post as documentation, but ultimately realized it's better suited as a post and I can do a much better job of documentation within the project.)

This structure of this blog is almost completely maintained by hand: I manually add entries to the index page, often going back to edit existing posts, and several entries are perennial – such as my Book list, as well as the syntopical collections on different sets of books.

That makes it fairly tricky to automate an RSS feed: at the same time, enough (4) people have directly reached out to request one. Instead of spending a lot of time building automation that will will inevitably frustrate me, I've decided to maintain it by hand instead (with ox-rss for exports). That means that there aren't going to be any entries for my older posts unless I happen to go back and significantly rewrite & update them.

If you happen to be new – welcome! – and I hope you find the content worth your time; and I'd love any feedback. A semi-curated selection:

On Emacs

• My writing setup
• Using Tmux with Emacs
• Building poet, my Emacs theme
• The configuration behind the blog

On Software engineering

• Metacognition on Debugging
• Always! include units
• A style guide for computational notebooks
• Using Python's settrace

On Books

• Almost everything I've read for the past several years
  • Favorites from 2016
  • Favorites from 2017
• [Extremely rough] Books on thinking

Drop me a note by email or Twitter in case you have any advice, comments or feedback!