Comments on: How (and why) we rewrote our production C++ frontend infrastructure in Rust https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/ A blog from the staff at NearlyFreeSpeech.NET. Thu, 14 May 2026 02:31:59 +0000 hourly 1 By: Douglas Muth https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/#comment-30511 Thu, 14 May 2026 02:31:59 +0000 https://blog.nearlyfreespeech.net/?p=866#comment-30511 It’s always good to read about a successful project like this, and I’m glad that after all these years, y’all are still building new things.

I can’t wait to see what features are rolled out for the service next!

]]> By: Onion https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/#comment-30503 Thu, 07 May 2026 01:03:45 +0000 https://blog.nearlyfreespeech.net/?p=866#comment-30503 Well that explains why I have to learn a bajillion new reasons “why” something exists when I try to learn something new with this infra. But constraints force creativity, and creativity forces learning. So at the end of the day we should all be grateful when things cause friction and force us to learn new paths. Thanks, homie.

]]> By: bmh https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/#comment-30482 Wed, 29 Apr 2026 19:48:34 +0000 https://blog.nearlyfreespeech.net/?p=866#comment-30482 Great work! Happy to not have noticed. 🙂

I think the biggest problem with AI vibe coding is delegation: it fragments cognition across multiple entities (you and the AI), which limits confidence in correctness. This can happen with delegating to a person, too, but it’s especially problematic with AI because of scale. The point is when systems are critical, one single mind should be both in charge and deep in the weeds.

Glad to know this wasn’t vibe-coded. 🙂

]]> By: jdw https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/#comment-30480 Tue, 28 Apr 2026 01:39:32 +0000 https://blog.nearlyfreespeech.net/?p=866#comment-30480 In reply to Nick Levinson.

All other problems with AI aside, just the idea of vibe coding a project like that makes me nauseous. -jdw

]]> By: Nick Levinson https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/#comment-30479 Sat, 25 Apr 2026 20:09:37 +0000 https://blog.nearlyfreespeech.net/?p=866#comment-30479 Sounds terrific. This description is why I don’t think AI is ready for prime time. I doubt anyone should’ve trusted AI to generate all that, including the testing, without some future major failure.

]]> By: Tim McCormack https://blog.nearlyfreespeech.net/2026/04/17/how-and-why-we-rewrote-our-production-c-frontend-infrastructure-in-rust/#comment-30473 Fri, 17 Apr 2026 21:44:44 +0000 https://blog.nearlyfreespeech.net/?p=866#comment-30473 Last year I rewrote a small but excitingly dangerous web service and we did something similar to your proxy testing, although in our case the switch, comparison, and telemetry code was built directly into the client.

I can confirm that you can find some really weird bugs (including existing behavior) that way. 🙂 Can’t recommend it enough.

]]>