Toots for burakemir@discuss.systems account

Written by burakemir on 2025-01-22 at 20:13

Bounties for Rust interop bindings and improving unsafe #Rust !

https://bughunters.google.com/blog/5273064917827584/level-up-your-open-source-karma-and-your-wallet-by-improving-security

=> More informations about this toot | View the thread

Written by burakemir on 2025-01-15 at 15:18

A propos of nothing, I was remembering this article, which tells a story of how unsoundness may involve unsafe #Rust but the bug can come from changes in safe code: https://notgull.net/cautionary-unsafe-tale/ - worth keeping in mind.

=> More informations about this toot | View the thread

Written by burakemir on 2025-01-08 at 15:15

Me, finding a python coding game for my bored 10yo girl

10yo: Daddyyy, coukd I not use #Rust instead?

They grow up so fast 😅

=> More informations about this toot | View the thread

Written by burakemir on 2025-01-04 at 06:43

Here is an attempt to explain #Rust lifetime parameters in few sentences.

Every time you pass around references (pointers) instead of the object itself, you want to make sure that the reference is valid. Meaning: the thing you point to (the referent) is not freed and is not moved or mutated when you access it.

The compiler does not know how to guarantee this. It does know when it does things to an object which invalidate references. Variables go out of scope, objects get moved or mutated which may cause moves. So it provides a sort of token of validity.

"Lifetime" parameters 'a are placeholders for such tokens. We use them to explain to the compiler that references are valid.

The tokens are a static concept, they come from program source and we never see them. All we need to know is that when a reference is guaranteed to be valid, the compiler can produce such a token. In using the parameters, we connect the facility that provides the tokens with the part of the compiler that checks validity requirements.

I hope something like this catches on because "lifetime parameter" remains very confusing as a name, and "set of program points" is maybe too much detail.

=> More informations about this toot | View the thread

Written by burakemir on 2024-12-22 at 17:42

[#]concurrency #async I sometimes tell this story of how async made it into programming languages over dinner. Anyone who wants to fact check or debate me on this? Little thread.

=> More informations about this toot | View the thread

Written by burakemir on 2024-12-20 at 08:33

"It is time to beat turnstiles into technology"

Henry G. Baker (1995) “Use-once” variables and linear objects: storage management, reflection and multi-threading https://dl.acm.org/doi/10.1145/199818.199860

Great article that has not lost its force 30 years later, talking about linear type discipline for managing memory and resources... all a few decades before it has become mainstream with #Rust.

=> More informations about this toot | View the thread

Written by burakemir on 2024-12-07 at 09:43

Personally, I am really not one for advent-of-anything as long as my project and side projects are in this state.

However, people might be interested in this one: https://www.rustfinity.com/advent-of-rust #rust

=> More informations about this toot | View the thread

Written by burakemir on 2024-11-03 at 21:10

Good story from the unsafe #Rust trenches on building a structure-of-arrays API https://tim-harding.github.io/blog/soa-rs/

=> More informations about this toot | View the thread

Written by burakemir on 2024-10-06 at 20:04

matklad's post is great: it may not offer fundamental new insights but it very succinctly articulates Rust advantages and areas that need work (interoperability) if we hope to see a high level language combined with one that permits fine-grained memory management. #rust

https://matklad.github.io/2024/10/06/ousterhouts-dichotomy.html

=> More informations about this toot | View the thread

Written by burakemir on 2024-09-20 at 19:14

Yesterday discovered my starlark parser (don't ask) had memory leaks, due to my careless use of heap allocated types in my bumpalo allocated ASTs (and not running with a sanitizer). This morning fixed all these in 1h. sanitizer, change types, bump allocate what I missed, step through problems in the IDE (vscode & rust-analyzer) done. I almost forgot that coding could be productive like that.

Do enable those sanitizers... #rust

=> More informations about this toot | View the thread

=> This profile with reblog | Go to burakemir@discuss.systems account