for someone who thought nolan was generally overrated until I watched oppenheimer — a truly great movie — is the odyssey good
rain 🌦️
@sunshowers.io
(gender)queer transsexual / they/them / eng @oxide.computer / atheist and scientific humanist / 🦀 #rustlang / #jj-vcs #1 fan / https://nexte.st / https://sunshowers.io also on mastodon: @rain@hachyderm.io / https://hachyderm.io/@rain
people stop paying attention to my writing midway because of the difficulty of my material 😭
Gergely Orosz@gergely.pragmaticengineer.com · 3d agoOne of the easiest ways to stand out when writing or creating videos or posting: DONT. USE. AI. Suddenly you sound distinctively different from 80%+ ppl who do and output the same verbose + nicely formatted stuff that feels all the same (+ most ppl stop paying attention midway)
the thing I hate most about LLMs is how using them properly requires constant vigilance. the slop-call beckons
Btw Claude doesn't have feelings but it does do this. Equivocates and adds qualifiers to everything
Katie Tightpussy, Editor-In-Chief@katie.bzky.team · 5d agosharing a personal issue that everyone i've told has said "Oww": the slippery behavior of weasel-wording and adding unnecessary qualifiers, especially when asked direct questions so as to construct a nebula of ambiguity around yourself, is a survival tool that breeds untrustworthiness when habitual.
I'd just like to interject for a moment. What you're referring to as an OpenAPI spec, is in fact, an API document, or as I've recently taken to calling it, an OpenAPI spec plus document. An OpenAPI spec is not a document unto itself but rather another free component of a fully functioning API system
Reminder that Chesterton's fence is basically the one good conservative idea
rain 🌦️@sunshowers.io · 6d agoif you were designing a stacked workflow from first principles, you would focus on atomic/bisectable commits as the unit of work, with interdiffs between versions to guide you. Basically what Gerrit, Phab and numerous other tools do. The lack of engagement with prior art continues to be frustrating
my favourite Michael Kalecki quote: “I have found out what economics is; it is the science of confusing stocks with flows”.
JW Mason@jwmason.bsky.social · 6mo agoPart of the problem is that we use the same word, "housing," to refer both to the flow of services and to the asset that produces those services. As if the word "ketchup" meant both the stuff we spread on burgers, and the factories that produce it.
great choice by rust to rename "object safety" to the much more intuitive "dyn compatibility (previously known as object safety)"
spent some time looking at this last week and was very unimpressed. stacked diffs/patch series don't make nearly as much sense with a new commit + merge workflow as they do with an amend + rebase workflow. jj compat is alao dubious. going to stick with spr
GitHub@github.com · 6d agoStacked PRs, now on GitHub. Go forth and stack 🥞
fascinating that the rust standard library also happened to run into soundness issues with allocators unwinding. (This is something I mentioned in my iddqd blog post as an entirely unexpected source of soundness issues that GPT flagged — it is now carefully handled by iddqd)
Guarantee `handle_alloc_error` never unwinds · Issue #834 · rust-lang/libs-team
Proposal Problem statement Currently, handle_alloc_error cannot unwind on stable. However, the unstable set_alloc_error_hook API would allow that to change, potentially allowing an unwind to happen...
github.com
btw iddqd is mythos approved 👍 it found one DoS pattern where while deserializing from serde, we would preallocate based on an untrusted size_hint
We're joining @anthropic.com's Project Glasswing, a collaborative effort to secure the world's most critical software. We're using Claude Mythos 5 to find and patch potential vulnerabilities across our codebase, from firmware through the OS, and control plane.
Oxide joins Anthropic's Project Glasswing
Using Claude Mythos 5 to protect critical software
buff.ly
fun fact, the belief within meta's infrastructure division is that if an architecture lasts 18 months, it is a success
Gergely Orosz@gergely.pragmaticengineer.com · last wk.Meta's performance review culture is becoming very problematic for its engineering culture. I said this a month ago in @pragmaticengineer.com (screenshot #1), and Semi Analysis said the same last week (screenshot #2). Engineers at Meta are world-class at optimizing for PSC...
claude with my review prompt is just way better at code reviews than I am. in particular, it is very good at finding things that aren't covered -- missed test coverage, comments, docs, etc. it'll sometimes make incorrect assumptions but it's almost always thought-provoking
throwback to the time I went to the Tate Modern and the words “Piet Mondrian’s ‘loss’” washed over me like a vision of God in all Their glory
Had a few losses lately, but this run annihilated the act 3 bosses -- zero damage from Aeonglass. Double-played Afterimage+ turn 1 is so broken
given that the peripheral is probably the worst show I've ever seen, not very optimistic for it
C. Robert Cargill@crobertcargill.bsky.social · 2w ago
I was maybe 2 years into living in the US, and very cis-presenting. Some coworkers were hanging out talking about bemani games. I walk up and bring up sdvx — they straight up don't believe me. I pull out my e-amusement card and hand it to them, and they pass it around, unable to believe it's real
My book officially has a cover! I'm so excited to watch it materialize. Next stop, pre-orders!
One of the things I'm used to doing is spending many hours painstakingly writing out a series of structs and enums which basically describe a spec and handle every possible case. I'm excited that LLMs make it possible to do this kind of work within the time constraints most devs face
github.com
guaranteed 50k because of the events in the next question mark room it makes possible