hl - ast and powerful log viewer and processor that converts JSON or logfmt logs Colors and conversion from other formats are helpful when taking a closer look at logs at the commandline, but filter capabilities are the most important feature.

I started following Lucene more closely again, because there are so many improvements to learn from. Today, using different data structures in case of segments with sparse deletions when tracking deleted documents, that yield a major memory reduction.

Have your Iceberg Cubed, Not Sorted: Meet Qbeast, the OTree Spatial Index TIL about OTree, a novel multidimensional index with efficient data sampling, suitable for distributed data stores apparently. The article also links to the corresponding paper.

polianna - a Java library/agent allowing to push JVM metrics via OTEL Also allows for native memory tracking, heap allocation rate, direct buffers and others.

kimojo-rs: A thread-per-core Linux io_uring async runtime for Rust optimized for latency Coming from azure, so I assume it's used in some modern data store, but apparently I missed the announcement.

Why On-Prem Kubernetes is Hard Nice article about the complexities and additional workload of running kubernetes yourself instead of using a cloud k8s service. First time I heard about "Day-2 operations", but I guess I am just hopelessly out of date once again.

Building a Durable Execution Engine With SQLite Great article with accompanying Java code, that uses annotations, bytebuddy and sqlite to be able to resume jobs. Great to follow explanations plus code.

Using ADBC in Java JNI bindings for for the C++ ADBC driver. Not yet fully feature complete though.

Two KubeCons, One Conference: While Everyone Demos AI Agents, Engineers Are Fighting With Syslogs I think this is one of the reason I have not been at many conferences, but rather to barcamps in the last years - you hear the status quo.

Dense Retrieval at Vinted: Nice blog introducing dense retrieval into a keyword driven search, starts with low recall sessions, a fine tuned CLIP model. retry strategies to switch between ANN and exact nearest neighbour search. Gets more interesting to the end.

Grebedoc: github pages for as Open Source Supports codeberg, uses Caddy, supports own domains, but not required.

How I fell in love with calendar.txt TIL about calendar.txt. The trend to text-only (and versionable) continues. Seems that many other TODO/calendar tools are so convoluted nowadays, that simplification is key.

A deep dive into BPF LPM trie performance and optimization What a great read about a problem (sometimes you don't need a solution just yet) and how it was found and debug, including some data structure refresher.

Finished: Latency Dives into the various concepts/ideas of reducing delay in software. More geared towards junior/mid level engineers who haven't had to touch latency/performance topics yet and want to get familiar with, explaining a lot of concepts. Around 240 pages, nice rust based code samples.

A hypothetical search engine on S3 with Tantivy and warm cache on NVMe Great post describing the required steps to get a simple S3 based implementation of a search engine. There's dozens of more gotchas, but the basics are most important.

Finally free of crutches, thanks to 3d printing I could at easily reach them after having a shower. Measure, create design, print prototype. Took half a day. And code in a git repo. Need to learn rounding edges, but we'll get there. Hoping no tools for post surgery equipment are involved.

LLM for Java weekend: embabel - Agent framework for the JVM Seems that quite a few early spring folks are involved in this one.

LLM for Java weekend: agent-o-rama An end-to-end LLM agent platform for Java and Clojure for building, tracing, testing, and monitoring agents with integrated storage and one-click deployment. Inspired by LangGraph/LangSmith.

Finished: Modern concurrency in Java Nice 300 page introduction into Virtual Threads and Structured Concurrency, how those two work together. As I didn't have time to delve into both topics so far, this was a great refresher.

After a year of work, we are happy to share that Spring Framework 7.0 GA is now available! spring.io/blog/2025/11... Time to celebrate, get a bit of rest and help the team to prepare Spring Boot 4.0 GA release planned November 20th 😊 #spring #java

Another day another query language. Querying JSON with JSON query language, making the queries look like JSON arrays. Might make sense, if all you have is JSON. The question is, if that is the right way of querying or just a backing syntax for an UI editor.

oxdraw - diagrams as code using mermaid but with adding the ability for exact positioning/coloring. Nice idea, also there is typst integration.

Super interesting post about fly.io state management software.Not a big consensus driven distributed system or centralized, but using SWIM and sqlite with CRDTs. Great to learn to see such design decisions play out on such a big network. fly.io/blog/corrosi...

How fast is java? Teaching an old dog new tricks Nice blog post about Java Vector API to improve rendering using SIMD.

Homerow - Keyboard shortcuts for every button in macOS successor to vimac, bringing vimium to osx. Is anyone using this? What is the difference to shortcat?

Achievement unlocked: Added a cover slide using a terminal with figlet to my no-slides #elasticon presentation showing ES|QL to look into three years of solar data. A part of the demo has also been written down in a blog post. See spinscale.de/posts/2025-1...

🛠️ If your team has issues to work through: We do an exercise at offsites called Elephants, Tigers & Paper Tigers. - Elephants are things that the group isn’t talking about but needs to - Tigers are things threatening the team, risks - Paper tigers are things that seem like risks, but aren’t

Introducing Chronos-2: From univariate to universal forecasting Reads like a major upgrade to the existing model.

Munich friends! I'll be in Munich for ElasticON Munich on Thursday. If you want to meet spontaneously on Wednesday noon/afternoon, let's have a coffee somewhere!

How We Saved $500,000 Per Year by Rolling Our Own “S3” Interesting very specific video processing use-case, where adapting to the actual workflow made more sense economically than using existing AWS infrastructure.

Katakate - Secure sandboxed compute for AI agents and workloads Self-hosted infra for lightweight VM sandboxes to safely execute untrusted code. Uses kata containers and firecracker (with jailer). Hadn't heard of kata containers before.

FeOS is a revolutionary init system for Linux, designed specifically for hypervisors and servers that run containers.

Javelit is a Java lightning fast data app development framework, inspired by Streamlit As a fan of streamlit and the way of prototyping small apps fast or having great demos, this sounds intriguing, and I'll test it as soon as I have a use-case.

Probably a decade late, but I finally started setting up neovim with a bit more effort. Using noise, telescope, checkmate for todos. Also added a gpg plugin, so that I encrypt files and share via dotfiles repo, while only being able to encrypt on personal systems. Other plugin recommendations?

Danube: a lightweight, cloud‑native messaging/pub-sub platform built in Rust Storing data in object storage, while using a WAL to keep latencies low. etcd is used for metadata storage. Uses Opendal to write to different object storages.

SierraDB: A Distributed Event Store Built in Rust Uses the redis protocol for communication. Neat trick to not need to support any clients out of the box - or at least skip the connection handling part, while having custom commands.

Performance Improvements in JDK 25 TLDR; always upgrade, free performance and auto vectorization

The AI-collapse pre-mortem Interesting piece about splitting the hype from the underlying technology and the generation it is expanding in. "... a collapsing economic bubble is not a useful data point about where this technology will end up."

Cloudflare sandbox SDK Couple of interesting features like code interpreter, allowing to run jupyter like snippets in a secure sandbox. Support for quickly doing git checkouts to support CI. Also allows for custom base docker images. Based on cloudflare containers

They gave an interesting talk at FOSDEM last year: archive.fosdem.org/2024/schedul...

Garage - an open-source distributed object storage service tailored for self-hosting S3 compatible data store, single binary, written in Rust, geo based replication, supposed to be faster than Minio. Not seen this so far, anyone running this in prod?

dysk - A CLI utility to get information on filesystems, like df but better Extending df: adds better formatting via table and colors to a CLI tool, looks instantly a lot more useful than df. Wish existing tools would not be considered 'finished' sometimes, but extended.

CPU Cache-Friendly Data Structures in Go: 10x Speed with Same Algorithm Nice examples for Go how to write CPU friendly code by not thrashing cache lines. skoredin.pro/blog/golang/...

Why TigerBeetle is the most interesting database in the world An article by an invested VC, but highly technical, with a lot of links to youtube talks and blog posts. Wondering if this is preparing for the series B here. Time will tell.

hygg - Minimalistic Vim-like TUI document reader Uses pandoc for document conversion, and then puts a TUI around the output neatly formatted, allowing easy selection, with vim keybindings.

rioterm - a hardware-accelerated GPU terminal emulator focusing to run in desktops and browsers Never heard of this one despite 5k github stars. Written in Rust just like wezterm, but also compiles to WASM (soon), and a focus on speed.

TIL there is GTFS - The general transit feed specification A specification for dynamic traffic data (delays, broken escalators, reroutes, everything around stations), in protobuf and heavily in production use - at least in Germany. You find realtime feeds, making it nice demo data.

arc - High-performance time-series data warehouse built on DuckDB, Parquet, and MinIO Configurable durability, automatic file compaction, uses DuckDB for querying, minio for storage, polars or arrow for parquet writing depending on the sent data.

openzl - A novel data compression framework By having custom codecs that allow to deal with the concrete features of your data openzl allows much higher compression ratios while still be fast on reading & writing.

Fundamentals of Virtual Memory Great post about memory features like paging, fragmentation, stack/heap allocation and more.