-
Switching to WezTerm
I switched to WezTerm on a 2012 ThinkPad T430 because it felt faster than GNOME Terminal and copy-paste was simpler. Benchmarks on the same machine show where that impression holds up, followed by the full Lua setup.
-
Review: Software Malpractice in the Age of AI
A review of Danny Tobey's argument that AI makes software malpractice harder to avoid, and why responsibility in software cannot be modeled exactly like medical malpractice.
-
[Book Review] Miner Town: Awakening (Miner Town, #1) by Ankit Saxena
An honest review of Miner Town Awakening, Ankit Saxena's ambitious dystopian science fiction debut - an imaginative world and a strong cast, held back a little by dense prose.
-
The 2D↔1D Problem in Text Editors
Editors show a 2D grid but store text as a 1D sequence. How textr, gedit, and Emacs bridge the gap, and why the choice follows the buffer's data structure.
-
Agent Loops as a Team Diagnostic
Four questions a team can ask about its own work, using Anthropic's agent loop taxonomy, to find the artifact it should write next.
-
JPMS and Cargo: Two Answers to the Same Problem
The same project built twice, in Java (JPMS + Maven) and Rust (Cargo) — how each answers compilation, distribution, and encapsulation, why a crate maps to a JPMS module and a Rust module to a Java package, and where the analogy breaks down.
-
Adopting Agent Loops in Order
The four agent loops depend on each other's artifacts. Notes on the order in which a team should adopt them.
-
What Sits Underneath the Agent Loops Post
Anthropic's Claude Code team described four kinds of agent loops. Notes on where the check on the work lives in each, and on the artifacts a team has to make durable underneath them.