claude research is kinda cool..booked 70% token (i used fable 5) and it researched < 10m- an extended web research pass- fetching primary sources (arXiv, USPTO patents, GitHub), vendor documentation (eg CodeRabbit), then crosschecking claims against each other before deep writeup
Shipping a GraalVM native image? -Dversion=1 at build time evaporates at runtime.Fix: read it in a static field + --initialize-at-build-time. GraalVM bakes the value into the binary at compile time. Same idea as Go's -ldflags "-X main.Version=1.0.0".
What can you learn about a Go codebase by ignoring semantics and vocabulary — and looking only at the shape of its functions?
Turns out: quite a lot. And also some things that look real but aren't. New writeup: https://t.co/IABXzUnu7l
Imagine finding a critical vulnerability in your OSS lib.
You patch it. You try to push to Maven Central.
You're capped.
That's the new reality after Aug 11, 2026. Here's what's changing. https://t.co/UGfvVVwRcC
hmm interesting.. given a specific java problem that only arises when using a native image( built via graalvm) .. claude 4.6 really really suffers to fix it, keeps doing the same thing or crazy solution.
the same issue presented to sonnet, it did it one shot. :O
Every BadgerDB app I've seen has the same 20 lines of encode/txn/decode boilerplate copy-pasted everywhere.
I got tired of it and wrote badgerx.
https://t.co/KFXJfdKlwR
Before → After 👇
#golang#badgerdb
Every BadgerDB app I've seen has the same 20 lines of encode/txn/decode boilerplate copy-pasted everywhere.
I got tired of it and wrote badgerx.
https://t.co/KFXJfdKlwR
Before → After 👇
#golang#badgerdb
I was recently intrigued by - does vocabulary-independent clustering find real patterns in real codebases.
https://t.co/wVLiM60nyl
Repo https://t.co/gkfO1qgwle
That rush to use up all your claude credits before the reset window is no joke .. i only got 2 mo mins to go ( and yes i am out here at the same time :P ) #furioustype#claude