Infrastructure

via Bob 🇨🇦🇲🇽🇺🇦 (@bielsubob@infosec.exchange): Canada built 44 bridges and tunnels across a single stretch of highway so that animals could cross without dying, and collisions with wildlife there have fallen by more than 80%, with grizzly mothers now teaching their cubs to use the crossings like a route the family has always known

Banff National Park’s 44 wildlife crossings — six vegetated overpasses and 38 underpasses, paired with fencing along the Trans-Canada Highway — have cut wildlife-vehicle collisions by more than 80% (96% for elk and deer). What I find most interesting is the multi-decade feedback loop: 250,000+ camera-logged crossings show animals learning the routes and passing that knowledge to their offspring, so the infrastructure gets more effective over time.

An interactive explainer on the physics behind GPS, covering how satellite signals, timing, and relativity all come together to pinpoint your location. A beautifully crafted piece that makes complex physics accessible and engaging.

via Rust Weekly 🦀 (@rust_discussions@mastodon.social): Introduction to durable execution with Flawless and Rust

Flawless is a durable execution engine for Rust: workflows that run to completion even when the process is killed mid-flight. Instead of hand-rolling a state machine with resume rules and idempotence guarantees around every external API call, you write plain business logic and let the engine handle the replay. Currently in beta 3.

via Thomas (@wheezle@hachyderm.io): Been writing code in the sans-IO style for a while now and really enjoying it so I wrote up a blog post about it! #rustlang

Firezone’s Thomas Eizinger explains the sans-IO pattern: implementing network protocols as pure state machines that never touch a socket or call Instant::now themselves, leaving all IO and timing to the caller. The payoff is fast, exhaustive, deterministic tests and a way to sidestep the “function colouring” problem — async ends up confined to a thin outer layer instead of infecting the whole stack.

via Rust Weekly 🦀 (@rust_discussions@mastodon.social): Introducing emit: developer-first diagnostics for Rust applications

Ashley Mannix introduces emit, a diagnostics framework for Rust that’s been years in the making. Rather than following OpenTelemetry’s data model, everything is a single primitive — an event, described by a message template — from which logs, trace spans, and metric samples are all built. It still plugs into OTLP collectors, rolling files, or a pretty-printed console when you want them.

A Changelog & Friends episode on self-hosted media servers — Plex, Jellyfin, Emby and the rest — and why keeping your own library still beats renting from streaming services. A good listen if you’re thinking about reclaiming your media from the cloud.

via Ethan Sholly (@shollyethan@fosstodon.org): Today I’m launching a project I’ve been working on for several months - a browsable directory of self-hosted software:

Ethan Sholly of the This Week in Self-Hosted newsletter has published his entire database of self-hosted software projects as a browsable directory. You can filter by tags and by the commercial services they replace, and sort by repo stars, recent development activity, or name — with the whole thing refreshed daily from the database he maintains for the newsletter.

via Nic Wise (@fastchicken@mastodon.nz): @timbray@cosocial.ca Finally found time to write up how I use Plex + Plexamp. Thanks for asking (the world) and prompting me to actually write the words

A practical guide to self-hosting your music library with Plex and PlexAmp as an alternative to Spotify, Apple Music, and other streaming services. Resonates with anyone who has accumulated a large MP3 collection over the years and wants to maintain control over music they’ve actually paid for.

via Simon Willison (@simon@simonwillison.net): And Fly.io announced their object storage alternative to S3! https://fly.io/blog/tigris-public-beta/

Fly.io introduces Tigris, a globally distributed, S3-compatible object storage service that caches objects near where they’re requested instead of pinning them to a single region. Fly’s take is refreshingly honest: they couldn’t figure out how to beat S3 themselves, so they’re pointing at someone who did — at least for the kinds of apps that run on Fly.

via Jan (@janpio@hachyderm.io): @scouten@ericscouten.social we have been trying https://codspeed.io/ for that use case. Works decently. Check github/prisma/prisma-engines PRs for examples.

CodSpeed runs benchmarks as part of CI and flags performance regressions directly in the pull request, using instrumentation rather than wall-clock timing so results aren’t drowned out by noisy shared runners. It supports Rust, Python, Node, and Go, and several open source projects use it to keep an eye on performance drift over time.

via Rust Weekly 🦀 (@rust_discussions@mastodon.social): Announcing Continuous Memory Profiling for Rust

Polar Signals, in collaboration with Materialize, has released rust-jemalloc-pprof, a library that exports jemalloc’s heap profiling capabilities in pprof format. This makes it much easier to debug memory leaks and understand heap usage in Rust applications, either as one-off investigations or as continuous profiling over time.

via Mikah Sargent (@mikahsargent@mastodon.social): 🤯 Blown away by this video from @film_girl showing the racks that house the tech for GitHub’s new M1 macOS Runners. Was not expecting to see the Minis deconstructed!

GitHub announces new custom M1 macOS runners for GitHub Actions, discussed in a sneak peek video with Christina Warren. This is a welcome addition for teams that need ARM-native macOS CI/CD pipelines.

via The Changelog podcast: From Docker to Dagger

Sam Ruby on Fly.io Blog: Multiple Logs for Resiliency

via Tim Bray: Apparently this just decloaked:

via Rust Weekly: Announcing mailfred 0.1, a library to offer services through the email infrastructure processing emails as requests & responses.