LRT: Async in WASI 0.3 externalizes the runtime entirely, making it possible to provide first-class async at the ABI level.
This means there is no runtime to thread through the program: the host responsible for running the Wasm program is the reactor.
All you do at the boundaries of your program is say things like: "this is a stream of values" or "this is a future", and the host + bindings generator will handle it correctly. This enables compositions of programs to share one runtime.
What's also neat is that for API abstractions on top of 0.3 syscalls, supporting sync and async requires roughly the same amount of work.
All you have to do is tell the bindings generator you want the async/.await version of the API, and the host just has support for that. Wrapping that in e.g. the Rust stdlib should be no extra work at all, given we had a place to expose the types from.
Legit question: why do data centers use vast amounts of water? It’s for cooling right?
I’m wondering why it isn’t possible to use heat pumps — potentially even at scale, to achieve cooling? I’m legitimately interested in why water is being used, and why alternatives are hard to deploy.
I know “heat” as a resource has a bunch of useful applications. Can be stored for later use, etc. Here in Copenhagen my home is heated by municipal heating. I’m wondering what makes this hard to apply to data centers?
Yay! The in-progress wasi:i2c proposal (WASI spec for embedded devices) just moved to phase 2!
This means it's not just something promising anymore, but it seems ready to be implemented in host systems and begin gathering implementation and user feedback from.
I'm legit very excited about this! — But this seems like it has a real shot at improving the embedded development experience. Virtual platform layering, local platform emulation, standard APIs, etc. etc. I'm into it!
@whitequark That was voted on in the WASI W3C sub-group meeting today! If you’re interested in attending those meetings — I believe they’re open, though I think you might need to sign up somewhere?
Never ceases to stump me that we have the technology to kill 99.95% of airborne viruses, proven to work, non-invasive, cheap to deploy and install, and is produced at scale already and we just like - collectively - kind of just don't really use it.
HEPA-grade air filtration is proven, cheap, and makes everyone's lives strictly better. It, like, makes zero fiscal sense for governments not to mandate its use in all covered public spaces ASAP.
@WagesOf ah right, yeah I agree that perfect should be the enemy of good.
Perfect filtration sure takes effort, but the point is not perfection — it’s reduction. Like, I can go to IKEA today and buy a HEPA-grade filter for $6. Not best in class, but certainly affordable. Why? Because they’re produced at scale.
“To make any dent at all” is unsubstantiated by evidence. It’s clear that any amount of ventilation is better than none. Any amount of filtration is better than none.
The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) has published standards for Indoor Air Quality (ASHRAE 62.1 & 62.2) and Control of Infectious Aerosols (ASHRAE 241).
I've done a first pass over the numbers published in ASHRAE 241, converting the cited numbers for various scenarios to cubic meters of clean air per person per hour. Then compared that to some of the air filters I've been looking at recently:
I'm looking into this because a number of pals are trying to create safer community spaces. The information on how to do that is available, but unevenly distributed - so I figured that I might as well read up (:
I now know that one group will probably be happy with a Mini Mk.2 purifier as part of their fixed deployment. With additional mitigations needed for when they run events.
Hi friends, I'm looking for tooling recommendations for doing property testing and fuzz testing in: Ruby, JavaScript/TypeScript, Go, Rust.
In Python, I'd use Hypothesis and Atheris but what are the good tools in those other languages?
I know that AFL++ can be used with targets written in any language, and that's great, but I'm also interested in language-specific tooling, if/where it exists.
Especially interested in personal success — I can Google too but want to know what worked for you.
Wanted: a small, finite number of RSS feeds I can follow to make sure I don't miss new concert announcements in #seattle. So far, The Stranger's / EverOut's feed has been quite good, but there's still a lot of stuff at Nectar, Neumos, et al. that doesn't get listed there.
Oh that’s amazing! Whenever I’m in mass transit with my partner, wearing the 3M basically means we can’t hold a conversation.
Not like we enjoy lengthy conversations while in transit anyway; but the option to would be nice. I’m going to try and get one I think. Thanks for the info!
Someone should write a user-oriented WASI-based OS, and I’m not even kidding.
I want a virtual OS I can load onto any operating system, give a manifest file, and it’ll deploy an entire server environment. Think: stripped down Synology OS, but virtualizable, and only runs WASI programs.
Its key feature would be: it can automatically update itself and its applications whenever updates become available.
Once again not loving that I can’t just write applications for my Synology NAS without needing to go through the complicated process of packaging it up in a very specific architecture-dependent way.
My NAS isn’t powerful enough to run Docker containers either — Wasm Components really would be perfect. But we don’t have a way to auto-update them… yet.
That reminds me that a good number of Michelin-guide restaurants serve:
foix gras
gold leaf (why?)
caviar
wagyu
You can find these nearly everywhere, and they’re always boring as shit. It’s again using something which is inherently human and can be fun and exciting as a status symbol instead.
I know very little about data frames, but at a glance they remind me a lot of differential dataflow? How would you articulate the differences between the two systems?
@xgranade oh interesting, thank you for explaining! — To clarify: by “data varies” do you mean just the data contained within, or potentially also even the schema?
By “stage of data processing”, is a good way to interpret this that data frames might be most useful to arrive at a useful analysis, and differential dataflow is useful when you need to make that analysis perform well later on?