matze :BlobHajMlem:

Plays software and develops bass

matze :BlobHajMlem: boosted:
Guillaume Gomezimperio@toot.cat
2026-01-23

Starting tomorrow, you will be able (on linux without cross-compilation) to install and use the Rust GCC backend directly from rustup! To do so:

rustup component add rustc-codegen-gcc

Thanks a lot to Kobzol for all their work to making it a reality!

github.com/rust-lang/rust/pull

#rust #gcc

matze :BlobHajMlem:matze
2026-01-22

@laund @argv_minus_one @chmod777 no one disagrees that tokio is large and complex. But, the fact of the matter is that tokio is used where it makes less sense and tend to be used synonymously with "async Rust" which at the end makes people think that's why async Rust is bad and difficult. Second, can you show me some data that proves that crates are becoming less runtime-dependent? It's not my impression.

matze :BlobHajMlem:matze
2026-01-21

@argv_minus_one @chmod777 the decision that every Future spawned as a tokio task requires viral Send+Sync because of its underlying work stealing concurrency model. That's the root cause for the borrowing pain and subsequent Arc<Mutex<…>> or actor bandaids you see in virtually all tokio-based async programs.

matze :BlobHajMlem:matze
2026-01-21

@chmod777 yes I mean embassy. Alternatives? Writing manual state machines? Sounds masochistic but each to their own.

matze :BlobHajMlem:matze
2026-01-21

@chmod777 tokio's decision is giving it a bad rap. Have you tried embedded? It's a Godsend compared to writing manual state machinery in C.

matze :BlobHajMlem:matze
2026-01-21

My main gripe with AI for coding? Either use "free" models like Gemini and shit is either too slow to keep up with Rust ecosystem churn or shell out moneys for Claude which will iterate towards some fixpoint eventually.

matze :BlobHajMlem:matze
2026-01-10

Got rid of Instagram on my phone. It turned into a creepy ad fest lately and I'm probably just too old to feel engaged by reels and stories. With that said, it was a good source of knowing when artists go on tour or play close by. Anyone has a viable alternative? Bandsintown?

matze :BlobHajMlem:matze
2026-01-03

@julia @fasterthanlime well whatever, there are RFCs and there is little to no progress. And without those I doubt there is much incentive for reqwest/hyper to become less dependent on tokio. However that dependency could be clearly marked in the docs for reqwest, I fell for it myself some time in the past.

matze :BlobHajMlem:matze
2026-01-03

@julia @fasterthanlime I mentioned traits for which there are synchronous versions in the standard library (Iterator, Read, Write, std::thread::sleep is not a trait but you get the point, …).

matze :BlobHajMlem:matze
2026-01-03

@julia @fasterthanlime the glorious day comes when async timers, I/O, streaming etc. traits are part of std.

matze :BlobHajMlem:matze
2026-01-03

@andrewnez I catch myself wondering why people think services like dependabot are a panacea to vulnerabilities. Merging those PRs nilly willy is doing nothing but enabling xz incidents. Proper auditing dependencies and their updates is the only sane way to go. Unfortunately, nothing exists on a cross language level as far as I can see. And even tooling like cargo vet is a bit painful.

matze :BlobHajMlem:matze
2026-01-02

These festive days, I've been experimenting with a "pure" async-driven UI progress bar and spinner library for called strides. There is the widespread indicatif crate but IMHO it does not mesh that well with async programs. Of course there are issues to resolve (I/O within poll() 😱), APIs to clean and features to complete but I call it good enough to base my binary installer on it.

github.com/matze/strides
github.com/matze/binge/commit/

Multi-progress spinner tracking three different futures
matze :BlobHajMlem:matze
2025-12-27

@dsoft thanks for these kind words! It means a lot 🙏

matze :BlobHajMlem:matze
2025-12-26

@BecherMax du als Stadt-Insider und @lagedernation Hörer: warum ist Karlsruhe nicht auf Mastodon?

matze :BlobHajMlem:matze
2025-12-25

Here's a small Christmas present 🎅

wastebin 3.4.0 – a small and fast pastebin – is out! Mostly small UX and performance improvements but also a rewrite of the Dockerfile for easier (for me) builds of both x86_64 and aarch64 Docker images and re-architecture of the highlighting component to extract the new theme showcase tool. Check it out:

github.com/matze/wastebin/rele

matze :BlobHajMlem:matze
2025-11-19

@kahvierzehn nee, verstehst du falsch. 1) der Sicherheitsgewinn ist immer noch da, Code crasht vorhersehbar statt "irgendwas" zu tun. 2) Den Check gibt es, den müssten Cloudflare-Mitarbeiter halt einbauen. 3) das Problem ist Teil einer ganzen Kette, aber das wird gerne ausgeblendet.

matze :BlobHajMlem: boosted:
Typsttypst
2025-10-30

You made the 2nd fastest growing language on GitHub: The amount of Typst code on there more than doubled this year!

Screenshot of GitHub blog post with table showing that Typst is the 2nd fastest growing language with over 3600 users and 108% Year over Year growth.
matze :BlobHajMlem:matze
2025-09-16

@jdortiz sorry for the late reply. No, I don't agree with that. `.await` is just sugar that the compiler uses to transform it and the surrounding async fn or block into a state machine type that implements `Future` again. You can write `let foo = async { tokio::time::sleep(Duration::from_secs(1)).await };` and then poll `foo`.

matze :BlobHajMlem:matze
2025-09-12

@jdortiz you might rephrase the "an executor is required to await something". Tokio, embassy, smol etc. aren't any special code and have to deal within the same constraints as any other code. It's entirely possible to drive a root future by manually polling it.

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst