thom

🖤Heavy Souls 🖤

i can't believe they have to play again today.

I would try doc.rust-lang.org/nightly/std/... rather than boxing to avoid the box overhead. That said, I'm at work at the moment so I can't really investigate. 1000 chains in a row is certainly a degenerate case and there's a number of reasons that could perform poorly, just not for the cited reason

Sum uses fold: doc.rust-lang.org/nightly/src/..., which as linked in my other comment does this. If you're seeing rust being slower than the other languages here, it's more likely because of copying around the iterator object, which will be considerably larger in Rust. A profile might be useful.

default impl doesn't call next in a loop. It calls fold: doc.rust-lang.org/nightly/src/.... In the case of Chain, that has a custom impl that calls all the first iterator and then all the second iterator: doc.rust-lang.org/nightly/src/.... which avoids the issue (intentionally).

Here’s my opinion on the new typescript compiler:

You should write programs in the language you want to.

[being assigned a routine task that falls under my job description at my place of employment] what did I do to deserve this

it's a picture of a in-progress gundam model kit. only the upper torso, head, and arms are completed. it is of the rx78-2 (the original gundam). the plastic used is all semi-transparent. on the desk you can see part of the assembly instructions, and behind you can see some of the runners.

wip gunpla

yeah my interest in semiprofessional tech things is at an all time low honestly >_>

i also like the aspect of twitter where people follow me for tech stuff and then i almost never tweet about tech, which this lacks... for now

i dont really like mastodon or bluesky but i guess of the two this one seems slightly better?

i still had this tab open i guess. the thing i love the most about it is how people rationalize software bugs in their heads. like, parallel universe is certainly a way you can explain that particular bug...

you see this kid of stuff with lots of speedrunning, and i love it every time.

its so good

I'd just like to interject for a moment. What you're refering to as Skeets, is in fact, AT/Skeets, or as I've recently taken to calling it, AT plus Skeets. Skeets is not a platform unto itself, but rather another free component of a fully functioning AT system made useful by the DIDs, XRPC, and

Map keys dont, Object does though, although theres some specialness with array. I guess you mean JSON tho

(actually in reality, i would probably be very stressed out by the whole thing and dissappear from the public eye to handle it, but still)

damn oklab is everywhere now. wild to me that not only did it get added to css ~2 years after the blog post, its the default colorspace for interpolation. i mean it owns so thats dope, just its crazy how fast the adoption has been. id never be sad again if a blog of mine had that kinda response

narrator: it does

god does bluesky really let me enter newlines in this textbox, which it completely deletes when i post? :|

just spent half a day at work tracking down a bug in our build system, only to realize not only is it in an open source project i maintain, the bug is one i WONTFIXed several months ago. damn, my petard.

admittedly "hard to prove correct" is probably what i signed up for when I typed `use core::intrinsics::{atomic_load_unordered, assume};`

feeling the tension between an optimal impl and one that's slightly slower but simplifies the correctness proof (well, more of a correctness argument)it's easier for me to prove correctness of my code if only one thread ever performs some initialization, despite the init being (trivially) idempotent

it's cool that i get to use my own domain i suppose