A catastrophic accident has struck the Mars outpost. Communication with Earth is completely severed. Only a handful of survivors remain.

Marsettler is a simulation where players shape the story of these survivors. Your journey becomes a narrative—a story worth sharing.

What’s in v0.0.0

• Autonomous AI: Characters find iron mines, extract ore, and haul it to the warehouse on their own
• Behavior Tree System: The architecture powering AI decision-making
• Resource Management: Mines deplete and disappear when exhausted

Run Simulation

How we found a bug in Go’s arm64 compiler by Thea Heinen

A release? Infrastructure changes? The position of Mars?

Our investigation stalled for a while at this point – making guesses, testing guesses, trying to infer if the panic rate went up or down, or if nothing changed.

A reproducible crash with standard library only? This felt like conclusive evidence that our problem was a runtime bug.

• https://github.com/golang/go/commit/f7cc61e7d7f77521e073137c6045ba73f66ef902

Index

• Performance
• Concurrency
• Distributed Systems
• Generics
• Reiaibility
• Others

Performance

We tried Go’s experimental Green Tea garbage collector and it didn’t help performance by Zach Musgrave

Parsing Protobuf Like Never Before by mcyoung

Deep dive into a go binary by Jesús Espino

If you want to know how high performance systems written in Go were built, read VictoriaLogs: https://github.com/VictoriaMetrics/VictoriaLogs by Phuong Le

1. Custom bloom filters to reduce disk I/O and CPU cycles for redundant logs.
2. Column-oriented block storage for better compression ratio and cache locality.
3. Memory-mapped files with automatic pread fallback for zero-copy reads.
4. Lock-free object pools and arena allocators to reduce heap allocations.
5. Reflection-free JSON parsers for streaming at hundreds of MB/s per core.
6. Compile-time templates replace text/html parsing with near-printf speed.
7. Dictionary-aware Zstd compression balancing CPU cost against bandwidth savings.
8. Multi-core parallelism everywhere with adaptive concurrency limits.
9. Scatter-gather fan-out writes with adaptive concurrency for network saturation.
10. Fast hashing and lock-free randomization for minimal contention.

How Go 1.24’s Swiss Tables saved us hundreds of gigabytes by Nayef Ghattas

Finding performance problems by diffing two Go profiles by Zach Musgrave

Optimising and Visualising Go Tests Parallelism: Why more cores don’t speed up your Go tests

Leveraging benchstat Projections in Go Benchmark Analysis!

Benchmarks and performance testing

New unique package by Michael Knyszek

This work also led us to reexamine finalizers, resulting in another proposal for an easier-to-use and more efficient replacement for finalizers. With a hash function for comparable values on the way as well, the future of building memory-efficient caches in Go is bright!

In the AI Era, Focusing on Essentials Instead of JPA Complexity by 백명석 (Translated from Korean)

The bigger problem was that JPA’s relationship mapping feature was too good.

JPA is overkill.

The problem of domain logic becoming dependent on JPA

Investment

When do VCs start seriously thinking about exits? by Demo Day (Translated from Korean)

For VCs, exits are not about “greed” but a “structural inevitability.”

As a founder, you must always check the ‘fund maturity date’ of your VC.

Human Resources

Things to watch out for when expanding team size by Seunghoon Lee (Translated from Korean)

1. Speed (e.g., product update speed, new technology development/launch speed, etc.) must increase compared to the past. (The idea that speed can slow down for the sake of stability and structure is 99% wrong. You must add stability and structure while making things even faster.)

News

November 2025

I Believe Rocket Lab Could Lose Half Its Value, Buy It Then by David H. Lerner

Summary

• Rocket Lab is a high-potential space company led by visionary CEO Peter Beck, expanding into defense and satellite markets.
• RKLB posted record Q2 revenue of $144M, up 36% YoY, and is executing well on contracts, including a $515M Space Development Agency project.
• The upcoming Neutron rocket launch is a major catalyst, but repeated delays and high expectations create significant short-term risk for the stock.
• RKLB is currently priced for perfection (P/S 52x); a failed Neutron launch could drive shares lower, presenting a potential buying opportunity.

How Netflix Uses Java - 2025 Edition by Paul Bakker

Impressive things.

• Virtual Threads
• Web MVC or WebFlux
• Comparison
  • GraphQL
    • Flexible schema to query data
    • Think in “data”, not in “methods”
  • gRPC
    • Highly performant Server-to-Server calls
    • Think “methods”, not “data”
  • REST 🪦

Using External Indexes, Metadata Stores, Catalogs and Caches to Accelerate Queries on Apache Parquet by Andrew Lamb (InfluxData)

Embedding User-Defined Indexes in Apache Parquet Files by Qi Zhu (Cloudera), Jigao Luo (Systems Group at TU Darmstadt), and Andrew Lamb (InfluxData)

Apache DataFusion 49.0.0 Released

Highlights I found impressive

• Dynamic Filters and TopK pushdown
• Async User-Defined Functions (ask_llm?!)
• Better Cancellation for Certain Long-Running Queries

I still remember learning to sail during my days as a midshipman. One moment stands out clearly: opening the sail and feeling the wind take hold, pushing the boat forward. Unlike rowing, where every stroke relies on your own strength, sailing carried me farther than I could ever go alone, lifted and driven by the power of the wind.

But the wind wasn’t always kind. Without warning, the boom swung around and hit me on the head. I was lucky to avoid serious injury, but the impact stayed with me. That was when I first realized: if you want to harness the wind, you need a helmet.

HYBE Bang Si-hyuk Scam Controversy: How Did He Pocket 400 Billion KRW? by Kim Soo-heon