Merlin Clos

@rovarma Why?

Ah yes, Wikipedia. Guess which “performance analysis tool” is not on the “List of performance analysis tools” page and probably never will be.

@Neolectron @colinlienard Oui ok, genre s’il y avait des bugs produits par Zig dans les test alors ils ont été reproduit en Rust, et ça c’est con

@MerlinClos @colinlienard Mais si les tests sont en TS, il faut les run soit dans node soit dans bun, et si avant tu les runnais avec Bun, bah c'est pas bon... J'ai la flemme de check mais t'as capté

Ya des gens ils utilisent des trucs comme ça en prod où leur runtime il est vibe codé en une semaine d'un langage a l'autre.

@Neolectron Il me fait rire car il disait « nan mais c’est un test ya genre 90% de chance que ce sois pas utilisé en prod vous inquiétez pas », et on se retrouve là

@colinlienard @Neolectron Je sais pas combien ils ont de % de coverage mais on peut pas nier que c’est intelligent d’avoir fait ces test en TS et donc ne dépendent pas du langage du runtime

@Neolectron Après si les tests passent 🤡

@__Barmy @Neolectron Pareil

@Neolectron Je commence à considérer deno

@Neolectron J’aimais bien Bun mais la…. en plus j’adore Rust mais faut pas deconner

@martindonadieu @Neolectron Ouais mais bon ça suffit pas, comme d’hab les model qu’on aura dans 1 an détruiront ceux actuels

@Neolectron pour rappel il a access illimité a Claude et surrement a des model qu'on a pas access avant 3 mois.

@jitl @glcst How SQLite parser can be faster than sqlite-parser ? 😭

i ported SQLite's parser to JavaScript and it turned out faster than every other JS SQL parser i tested: 2.5x faster vs liteparser (wasm) 6x faster vs guanmingchiu/sqlparser-ts (wasm) 10x faster vs node-sql-parser 100x faster vs pgsql-ast-parser 200x faster vs sqlite-parser

@JesseStojan @Malix_Labs Okay Einstein

@Malix_Labs I'm having a hard time trying to see the logic in this when memory management is so simple that nobody I've worked with professionally has brought this up as an issue when using C or C++, it's just not difficult and I've never met someone who actually struggles with it.

C and Zig are only easier to read compared to Rust if you don't mind memory issues If you care about memory issues, they become, literally, the two hardest programming languages to read, both for you and for the AIs

Rust + Tokio → Async Systems Rust + Actix → High-Performance APIs Rust + Axum → Modern Backend Rust + Diesel → ORM / DB Access Rust + SQLx → Async Database Rust + WebAssembly → Browser Apps Rust + Yew → Frontend (WASM) Rust + Tauri → Desktop Apps Rust + egui → Native GUIs Rust + Bevy → Game Development Rust + wgpu → Graphics Rust + Embedded Rust → Firmware / IoT Rust + Linux Kernel → Systems Rust + Polars → Data Processing Rust + Arrow → Analytics one language >performance >safety >systems >full-stack (yes) and people still say “rust is too hard” 👇

@joncphillips Kk I understand, in fact I’m using MariaDB but that’s the same setup than PG. It’s just that I run all my services in Docker so they’re easy to set up, and if I used SQLite it would break the consistency of the management and increase coupling by having the DB in the back

@MerlinClos A separate process to monitor, a connection to manage, and one more thing that can go down independently of your app. If that's no big deal to you, use Postgres. Genuinely. The tweet was aimed at people who haven't shipped yet and are overthinking infrastructure.

Hot take: SQLite in production is fine for 90% of the SaaS products people are building. Your app doesn’t need Postgres.

@notreaIlyjohn @neogoose_btw Because with Rust, you know exactly where it's unsafe since you specify it :)

@neogoose_btw Why do this if you can program in C++ ?

here is one example when I specifically violating rust safety rules for performance fff guarantees that there are no allocations after the index is built, but there is no way in rust model to describe that a Vec is never going to be updated after certain point in time this little trick allows me to do updates to the mtime/size/git status without holding the locks at a cost of risk that it is going to blow me up and this is fine, at least I know WHERE the blow may happen

@CodeWithAmann The question isn't even worth asking, it's Go

Which backend stack has the strongest future?

@theodorebeers @neogoose_btw Makes sense but in that case, they could have simply delved deeper into the discussion instead of getting into this "beef"

@neogoose_btw In an interesting counterpoint, a member of the Zig project commented that they felt Zig as a language was judged negatively due to crashes in Bun. I think they would rather the language be judged by Ghostty and TigerBeetle.

When I mentioned that bun has more crashes than deno & node because of zig I got banned by Jarred. But finally people start getting that having compiler enforce your safety even in 95% of cases is actually amazing and it’s worth basically any suffering of a borrow checker.

@neogoose_btw Where did you get banned?

Unpopular opinion : Linux gets easy the moment you stop expecting it to behave like Windows

Crazyyyyy update, GJ @PapaTunkasina 🤝

@ledevnovice Moi

Qui fait des branches de feature sur ses projets perso les ami(e)s codeurs ?

No server components in Vue was intentional