posts
- Pulling plugs on purpose
The first power-bank test passed because a second supply was hiding the failure. What each test in the UPS project could prove, and what it couldn't.
- The guard rails that said no
A deploy script that aborted, a permission check that refused firmware, a password with one copy, and a grep that said yes when it should have failed.
- A Broken AOT Cache Still Serves 200
By default the JVM logs one error and keeps running when its AOT cache is broken, so a Leyden tier without AOTMode=on can measure plain java -jar.
- Empty diff, finished code: 124,000 characters of thinking
Two K2-Horizon agent runs scored 29 with an empty diff. The code was in the reasoning: the Go cell's, pasted into the repo, passes 27 of 27 hidden tests.
- The harness decided how long the model thought
One 27B model, one llama.cpp server: 250 s per agent run in OpenCode, 116 s in Pi. Under OpenCode the model reasoned 3.6 times as much.
- The alert about the blackout needs the internet too
Power-cut alerts leave over an uplink the cut can take down: retries, per-channel queues, a router watchdog, and a status LED that goes dark.
- No Manual Intervention Required
An LTS-to-LTS Quarkus bump, 3.33 to 3.40, changed no production Java and broke the tests twice: one class loaded twice, one database shared by six branches.
- Same Builder, Different Binary
One GraalVM builder image, one banner on every build log, 18 native binaries. Twelve came out PIE and six didn't, and nobody had touched a linker option.
- The Flag Was Verified. The File Was Never Read.
Spring's process-aot writes its own native-image.properties over yours. A verified GraalVM flag sat in the repo for 43 days and never reached the build.
- Health 200, API 200, Prometheus 000
A GraalVM 25i4 bump made native /prometheus stop answering. The reflection gap was old; a new default turned a missing metric into a hang.