Who contributed to Rails on May 2, 2026?

2 developers shipped this update, including Korri and mugitti9.

What were the notable Rails updates?

v4.5.0, Don't crash on client disconnect (Server and Application), and Defer ViewReloader build when no view paths are registered.

@rails

Ruby on Rails - Shopify, Basecamp, GitHub all run on it

github ↗

Pick a date

Topics: Ruby Full archive →

The Wire · Showcase

SPRING STOPS CRASHING ON SLOW BOOTS, RAILS PATCHES VIEW RELOADER

By RepoJournal · Filed 06:00 UTC on May 2, 2026 · About Rails

2 people shipped this

Korri @Korri 2 cited

mugitti9 @mugitti9 1 cited

Spring 4.5.0 fixes the Errno::EPIPE crashes that kill processes when clients disconnect during preload, a critical stability issue for large monorepos that take >5 seconds to boot.

The Spring release [1] addresses a years-old reliability problem: when a client times out waiting for the preloader to finish (default 5 second window), unrescued socket writes [2] crash both the server and application processes. Shopify's monorepo made this reproducible on every slow boot. The same release adds post-environment-load hooks [1] and eagerly preloads framework base classes before the parent process forks, preventing side effects during first request. Meanwhile, Rails merged a ViewReloader fix [3] that stops constructing empty FileUpdateCheckers on every engine boot — a performance win for apps with lazy view paths. The ecosystem also landed a minitest compatibility fix [4] for request format assertions that broke when minitest changed how assert_nil works under the hood.

Action items

→ Upgrade spring to 4.5.0 immediately if you see EPIPE crashes on slow boots rails/spring [immediate]
→ Update rails main branch to pick up ViewReloader optimization before next release rails/rails [monitor]
→ Verify request format tests pass with latest minitest if you override assert_nil rails/rails [plan]

References

[1] v4.5.0 ↗ rails/spring
[2] Don't crash on client disconnect (Server and Application) ↗ rails/spring
[3] Defer ViewReloader build when no view paths are registered ↗ rails/rails
[4] Avoid assert_nil for Mime::NullType request format ↗ rails/rails

Quick answers

What shipped in Rails on May 2, 2026?: Spring 4.5.0 fixes the Errno::EPIPE crashes that kill processes when clients disconnect during preload, a critical stability issue for large monorepos that take >5 seconds to boot. In total, 19 commits, 8 pull requests, and 1 releases landed.
Who contributed to Rails on May 2, 2026?: 2 developers shipped this update, including Korri and mugitti9.
What were the notable Rails updates?: v4.5.0, Don't crash on client disconnect (Server and Application), and Defer ViewReloader build when no view paths are registered.

RAILS 8 LAZY ROUTES NOW THREAD-SAFE, SOLID QUEUE ADDS MIGRATION GENERATOR

A race condition in Rails 8's deferred route loading could serve 404s on valid routes under concurrent requests, and a fix landed overnight.

ruby 18 shipped 1-min read

@rails 1 day ago

SOLID QUEUE 1.5.0 SHIPS WITH CRITICAL CONCURRENCY FIXES

Solid Queue v1.5.0 lands three separate fixes for a production bug that corrupts semaphore counts and lets concurrency-limited jobs run wild.

ruby 45 shipped 1-min read

@rails 2 days ago

Rails Strips Dead Code, Solid Queue Hardens Tests and Drops Ruby 3.1

Aaron Patterson is cleaning up Rails internals while Solid Queue fixes flaky CI and raises its minimum Ruby version to 3.2.

ruby 17 shipped 1-min read

@rails 3 days ago

RAILS CORE CLEANS HOUSE WITH FOUR DEPRECATION SWEEPS

Rails is systematically removing unused internals across Journey, RedisCacheStore, and test utilities, signaling a major cleanup push before the next major version.

ruby 74 shipped 1-min read

Elsewhere on the wire

AI Agents about 9 hours ago

CLAUDE OPUS 5 LANDS ACROSS THE STACK

The newest Anthropic model is now live in langchain, Cline, and llama-index, with native support for extended reasoning and 1M context windows.

ai-agents 28 shipped 1-min read

Local LLMs about 9 hours ago

OLLAMA LANDS LAGUNA SUPPORT AND CRUSHES MEMORY LEAKS WHILE SGLANG HITS V0.5.16 WITH CONFIDENCE-DRIVEN SPECULATIVE DECODING

Ollama shipped three critical performance and reliability fixes for Metal residency and concurrent access patterns, while SGL-Lang released 0.5.16 with a new speculative algorithm hitting 383.7 tok/s on DeepSeek-V4.