cuppa

brewed 03:03 AM
← May 30

Sunday

may31

2026

Jun 01 →

the brief

A practical day: AV2’s v1.0 spec lands, Next.js sharpens Turbopack in a fresh canary, and Simon Willison pushes Python ASGI apps fully into the browser. Research explores running large MoE models under tight VRAM, while Microsoft’s handling of perpetual Office stirs lock‑in concerns alongside thoughtful voices on sandbox transparency and stepping back from tech.

the poursit · sip · 7 items

alerts

(01)

pulse

(02)
  • hn/frontpage· feedMay 30, 09:46 PM

    AV2 video standard hits v1.0

    AOMedia finalized the AV2 v1.0 spec, giving developers a stable, royalty‑free next‑gen codec promising improved compression efficiency for web, mobile, and streaming pipelines.

    The AV2 Video Standard Has Released (Final v1.0 Specification) — Article URL: https://av2.aomedia.org Comments URL: https://news.ycombinator.com/item?id=48340910 Points: 35 # Comments: 1

    signal 5hype 1video_codecstandard_releaseav2source ↗
  • vercel/next.js· feedMay 31, 12:00 AM

    Next.js canary sharpens Turbopack analyzer

    v16.3.0‑canary.36 brings multiple Turbopack analyzer fixes and explainability improvements plus clearer CLI failures, tightening feedback loops for teams living on the bleeding edge.

    v16.3.0-canary.36 — Misc Changes Turbopack analyzer: analyze ternaries + promises in string & nullish methods: #94187 [turbopack] Break up graph.rs in the analyzer into smaller files: #94211 [turbopack] Add explain() methods to WellKnownObjectKind and WellKnownFunctionKind: #94217 [turbopack] Handle negative numbers as constants in the analyzer: #94218 [turbopack] Use ast_path.kinds() instead of mapping over ast_path in the analyzer: #94221 [CLI] Exit non-zero with a clear message when next t...

    signal 7hype 1nextjsturbopackrelease_notessource ↗

findings

(02)

voices

(02)
  • simonw/blog· feedMay 30, 09:36 PM

    How Anthropic contains Claude sandboxing

    Willison highlights Anthropic’s unusually thorough write‑up of sandboxing across Claude products—useful transparency that helps engineers reason about trust boundaries and containment guarantees.

    How we contain Claude across products — <p><strong><a href="https://www.anthropic.com/engineering/how-we-contain-claude">How we contain Claude across products</a></strong></p> A complaint I often have about sandboxing products is that they are rarely thoroughly <em>documented</em>, and in the absence of detailed documentation it's hard to know how much I can trust them.</p> <p>Anthropic just published a fantastic overview of how their various sandbox techniques work across <a href="https://cl...

    signal 9hype 1sandboxingsecurityanthropicsource ↗
  • simonw/blog· feedMay 30, 07:39 PM

    Retiring from tech to live offline

    Chad Whitacre’s typewritten letter describes a deliberate, values‑driven exit from tech—not AI panic—capturing a growing undercurrent of burnout and re‑prioritization in the industry.

    I Am Retiring from Tech to Live Offline — <p><strong><a href="https://openpath.quest/2026/i-am-retiring-from-tech-to-live-offline/">I Am Retiring from Tech to Live Offline</a></strong></p> I've seen a lot of posts on forums from people threatening to quit their careers over AI. This is <em>not</em> one of those: Chad Whitacre is taking concrete steps, starting with this typewritten, scanned letter</p> <blockquote> <p>I'm retiring from tech. Well, "retiring" is euphemistic. I'm stepping away f...