Field Notes
Notes on event-driven Java, Spring Boot, Kafka, and secure systems for regulated banking.
Two Claude Code mods, Flightdeck and a prompt-cache meter, tested with an 8-agent workflow on a Play/Scala repo. A mod doesn't change the model; it changes what you see while it works, and so what you decide.
Cats Core · Part 9
Scala's == compares any two values, so a TodoId against a String compiles and returns false. Cats' Eq makes that a type error, explains why Some(1) === None fails, and lets me decide which fields count as equal.
Guice and play and escaping DI hell · Part 2
A test that could not pass forced my first trait into the graph. @ImplementedBy, a root Module, a test that builds the real Guice graph, and the error Guice prints when a binding is missing.
Guice and play and escaping DI hell · Part 1
At work I wire Play apps by hand with the cake pattern. In my own projects I switched a Play 3 template to Guice: one deleted config line, one @Inject(), and a timestamp that made me think Guice was broken.
I built a small commit-message plugin and ran claude plugin eval on it. The no-plugin baseline showed what the plugin actually adds, and the one red case turned out to be the judge's mistake.
How this blog's search works end to end: a weighted tsvector side table kept by a trigger, a thin Anorm query with websearch_to_tsquery and ts_rank, and highlighted snippets split on control characters instead of rendered HTML.
Scala and the JVM · Part 20
A hand-written unapply turns a catch block into a policy: pass your own errors through, wrap recoverable JVM exceptions, and let fatal ones propagate. A runnable Scala 2.13 demo shows exactly when the matcher calls it.
Cats Core · Part 8
I wrote a tiny Todo validator with cats' ValidatedNec and it made one rule click: you can only accumulate errors when the steps don't depend on each other. That's the whole difference between an Applicative and a Monad.
Anthropic's Claude Code team review agent-written code in layers. I mapped their stack onto CV Next: an independent Claude review on every PR, a verdict tied to a commit, and a /ship skill that waits for it. Plus what I left out, and what broke on the first run.
I ran a real feature through Claude Code's opusplan setting, then had Opus deep-review what Sonnet built. The two worst findings were lines in the plan I had approved.