July 11, 2026
Crash now, argue forever
A Erlang style pure Scheme Webserver and further
This self-fixing web server wowed coders — then the comment section turned savage
TLDR: Igropyr is a new web server project that promises to keep running even when parts of it fail and to update itself without going offline. Commenters instantly split between impressed, deeply skeptical, and annoyed that older project history may have been erased under a shiny new launch.
A new project called Igropyr is pitching a very dramatic dream: a website server that doesn’t panic when things break. Instead, if a request crashes, it swaps in a fresh worker and tries again. It can also update parts of itself live, without rebooting the whole service, which is catnip for people who love the idea of changing a site while it’s still running. On paper, it’s a slick pitch: self-healing, live updates, and even a way to tell users exactly what kind of failure happened instead of just showing a vague error page.
But the real action was in the replies, where the applause quickly collided with eye-rolling. Creator guenchi dropped a feature roll call like a movie trailer — “Let It Crash, Hot code swapping” and more — but critics were not buying the hype. One commenter, mullr, went full scorched-earth, calling it “another 2 day old vibe coded project” and accusing the site of feeling slapped together. Ouch. Then came the nostalgia bomb: hellcow lamented that the project’s older history had been “completely nuked” to make way for the current version, turning a code launch into a mini culture-war about whether shiny new AI-era presentation is erasing real craftsmanship.
Not everyone came to throw tomatoes. edfsadfer coolly pointed out that Swish, a similar Scheme project, has been around for about a decade — a classic internet move that basically translates to: nice trick, kid, but the elders have receipts. So the vibe was split between “this is exciting”, “this is overhyped”, and “please respect your ancestors.” In other words: perfect comment-section chaos.
Key Points
- •Igropyr is presented as a pure Chez Scheme web server using Erlang-style actors and a libuv event loop.
- •The article says each request runs in a supervised worker pool where crashed workers are replaced and requests can be retried up to three times.
- •Workers that exceed a configured execution threshold can be killed and replaced so a stuck loop does not freeze the server.
- •Routes and handlers can be hot-swapped on a running server, with in-flight requests finishing on the old code and new requests using the updated code.
- •The server can return structured `crash` or `stuck` fault messages while keeping the client connection alive for retrying requests.