This is a pretty good post calling out Andrew Kelly (Zig) and Filip Pizło's (Fil-C) crank opinions about memory safety in Rust:
https://itsallaboutthebit.com/memory-safety-absolutists
Remote
yosh
@yosh@toot.yosh.is
mastodon 4.7.3Be kind to people, be ruthless to systems.
Concurrent Computing ←
Programming Language Design ←
WebAssembly and Rust ←
u(๑╹ᆺ╹)
3050 Followers
526 Following
50 Posts
Joined November 04, 2022
Pronouns:
He/They
Location:
København, Danmark (9F7JMHJM+28)
Timezone:
CEST (UTC+2)
`Infallible = !` in Rust is apparently now in FCP! I'm so excited for this!
On the Bytecode Alliance blog: The Road to Component Model 1.0
https://bytecodealliance.org/articles/the-road-to-component-model-1-0
(This post is from yesterday, WASI 0.3 release post will go up tomorrow)
Replying to
On the same topic - I'm also really annoyed by the “real GitHub Uptime tracker" that's regularly posted. Evan Hahn did a good job breaking down why it's bad: https://evanhahn.com/in-defense-of-githubs-poor-uptime/
The last paragraph in that post is titled: “There are better reasons to dislike GitHub”. And like, yeah - definitely.
Replying to
Of course there are points in the middle about the impunity of corporations, double standards, and the influence of money in legal systems. But that's not the critique I usually see.
I want the desire to protect public libraries to be larger than the desire to punish corporations.
Some decent progress today on my Wasm Component package manager! I managed to extract the metadata from components and started rendering it out.
This builds on some of the metadata work I did last year, adding a set of conventional custom sections to wasm-tools for things like license info, author data, and revisions.
"Share what you make, not what you hate" is still one of those things I tell myself regularly.
Yay, it’s now possible to call Wasm Components directly from the CLI using a familiar command/flags/args notation!
AI, math
Ohh yay, Wasmcloud just uploaded a number of their interfaces to wasm.directory!
https://wasm.directory/wasmcloud
This includes interfaces for Postgres, secret management, and messaging. Very cool!
Pet peeve in kitchen design YouTube: the misunderstanding of "the kitchen triangle". The distance between cooktop, sink, and fridge is not actually important (up to a limit ofc). There are three more important things to pay attention to:
1. The sink -> prep -> cooktop transition.
2. Having a fridge accessible from corridors so nobody needs to cross the active cooking zone to grab a snack or drink.
3. Having a load/unload zone close to the fridge and oven.
Replying to
One of these days I'm going to learn what the job of a "release manager" is and I suspect I might have an epiphany.
I feel like I need to tell the world about labneh. If you eat dairy: labneh is easily top 3 dairy products of all time for me.
Replying to
@ekuber@hachyderm.io Some people have a tiny cop living in their brain and they can't picture what it's like not having one.
You're allowed to just do stuff! Even if to some people that might seem foreign and unthinkable!
Replying to
Personally not a fan of introducing the concept of lookahead in execution ordering.
Whenever I hear anyone in positions of power say: "It travels through respiratory droplets" I'm now conditioned to believe that it also travels through aerosols as well. But that's scary to say, so "droplets" is the more palatable lie.
Are you excited for quieter construction work? I know I am!
Replying to
Replying to
@ceejbot@toot.cat
I think it's because people have been lauding the benefits of LLMs for about four years now, and it's only like in the last six months or so that they've actually become useful. And that's not even true across the board.
It's hard to tell truth from lie when the message is unchanged even though the technology has changed.
Replying to
@scrabsha@hachyderm.io
Do you have any thoughts about the interactions between view types and pattern types?
I don't care much about the final notation, as long as both features end up working well with each other. Because both features are really cool and I want both ^^
Replying to
@andrewnez not me knowing my package manager is susceptible to like, half of this list 😭
I've got stuff to go fix now heh
Replying to
@samir Ok, sure but like - what does that mean? Don't let people push commits beyond a limit? Halt the creation of new repos? Stop accepting new users?
One thing GitHub has recently done is stop accepting additional individual sign-ups to Copilot.
Replying to
@samir I'm not sure I'd trade limited (but regular) stability issues for commit and PR quotas personally. That... seems worse? Like a lot worse.
Boosted by @joe@f.duriansoftware.com
Replying to
@joe@f.duriansoftware.com
Perhaps @delta@chaos.social matches what you're thinking of? I believe keys are kept on-device / synced out-of-band, even though messages are stored on regular mail servers.
Replying to
✅ performance review submitted
when the Unifi travel router works, it *works*. It's so nice to just be able to sit here in a hotel room and have everything automatically connect ^^
Replying to
@tty@sunbeam.city They aren’t, no - droplets have a roughly 2m travel radius and eventually fall to the ground. Aerosols can travel much much further and stay airborne for hours.
The mitigations for both modes of transport are vastly different. Cloth/surgical masks work OK for droplets but do little for aerosols. It took the WHO more than 2 years to admit that Covid-19 was actually airborne, e.g. transmitted through aerosols. Which was far too little, far too late.
Replying to
Why do you believe that? This PR is not T-lang nominated and afaict landing this is a T-Libs-API + T-Types co-stabilization.
The broader stabilization of ! will involve lang, but I don't see why this particular PR would require lang sign-off?
Replying to
@dln I don't believe a single git commit neatly maps to a single RPC request so I'm not sure we can extrapolate "total commits handled" to "requests per second".
Replying to
@chrisdenton@hachyderm.io ahhh ok, yeah that changes things!
I'm surprised it isn't t lang nominated honestly
Replying to
Replying to
@amt@hachyderm.io oh yeah, good one. less important than the other 3 imo, but not unimportant either.
Replying to
@poliorcetics
Sorry, what are you disappointed about? The growth data is a plain fact. And yeah, I've seen systems become unstable with far less throughput increase than this.
Replying to
@jrose@social.belkadan.com wrote most of a blog post today, will probably publish tomorrow. But let's see heh.
Here's the intro, which I hope helps illustrate what I mean:
Replying to
@adredish@neuromatch.social I don't know, you tell me. Do you see an issue here with the attribution part, with the profit part, or with both?
I've got to brush up on hypervisor knowledge for work over the coming weeks. Please send me your favorite (introductory) resources, if you have any!
Replying to
@simon@tutut.delire.party yeah I agree, but where do we then want to go from here? I see two options:
- everyone gets punished
- nobody gets punished
I prefer the latter over the former. But I almost exclusively see people argue for the former.
Replying to
@poliorcetics I guess I'm mostly interested in steering this to something that to me feels more grounded. Like for example: I think it's fair to ask why GitHub can't handle the increase in volume.
But I get annoyed when I see dashboards that have a broken methodology or discount things like volume increases. Like, I want some of the critique to be… better I guess?
Replying to
@Fishrock@mastodon.social this isn't the full thing yet, but it's one of the biggest steps to actually get there — so it sure seems that way!
Replying to
@marmrt@mastodon.nu I've seen plenty of opinions about the IP rights of the New York Times and the Walt Disney corporation, so I don't think the commentary is necessarily just scoped to software.