ทวีตที่ปักหมุด
If you're stressed, burned out, and need to loosen up, consider building your own operating system. #rustlang
English
Tin Rabzelj
2.2K posts
Tin Rabzelj
@tinrab
Building cool things. Rust, TS/React, AI 🚀 https://t.co/I1M1b7gerz
Slovenia เข้าร่วม Ekim 2021
70 กำลังติดตาม312 ผู้ติดตาม
@yacinelearning These are all over my timeline. It definitely reads like it was LLM-generated. Idk, it's probably just regular engagement bait.
English
@ssijak Then they should just say they disagree instead of bringing up bad examples constantly for a week after 🙃
English
@theo Saying an LLM isn't good at "frontend" seems wrong. If a model is pretrained on a large enough coding dataset, you probably can't say they aren't good at some specific programming task. The issue is probably somewhere else, like the prompt.
English
This post was so bad that it made me crash out a bit. OpenAI models are not good at frontend. The examples in this article are embarrassing and I actually can't believe they posted it.
x.com/sherwinwu/stat…
Sherwin Wu@sherwinwu
If you put in the thought and intentionality required for it, you actually get pretty great frontends from GPT-5.4! Check out some best practices here: developers.openai.com/blog/designing…
English
I think it looks great, and it will only get better. We need to see more demos, though. No idea what kind of model it is (VAE? # of params? it works on color buffer and motion vectors, afaik). I think it makes sense to explore graphics in this direction. Can you go further with this tech than just increasing what gpus do with the regular rendering pipeline? You could compare this with increasing settings from "medium" to "epic". Would it be more efficient on the gpu? Idk. Also, you could just apply this "filter" with lesser intensity, so it wouldn't necessarily look completely AI-generated.
English
I didn't even consider DS2 but put side by side you can immediately see that something is just off about Grace, it's not like early AI, it's a much more subtle uncanny valley issue. Something about her skin just looks plasticity, the light doesn't scatter on her skin correctly
TheGameVerse@TheGameVerse
Death Stranding 2 Nvidia DLSS 5
English
@CloudflareDev You could do a funny business move by making this the only bot that bypasses CF's bot detection.
English
We want to clarify something: one of the motivations for building this feature is that building a crawler that respects crawling best practices is hard. Too many crawlers use this as an excuse not to abide by site owners' preferences, and respect directives like robots.txt.
/crawl does this out of the box.
Additionally, the /crawl endpoint cannot bypass Cloudflare's bot detection or Captchas, and self-identifies as a bot. It respects all the same protections available to operators today: AI Crawl Control, robots.txt, Content Signals and Pay-Per-Crawl.
So, site owners can and should still choose how their content is consumed, and our /crawl endpoint will respect it.
English
Introducing the new /crawl endpoint - one API call and an entire site crawled.
No scripts. No browser management. Just the content in HTML, Markdown, or JSON.
English
Devlog.
I'm working on implementing a vector database in Rust based on LSM-VEC paper (2025 arxiv.org/abs/2505.17152) and ACORN (2024 arxiv.org/abs/2403.04871).
It's very much WIP.
LSM-VEC uses LSM-tree storage with graph-based ANN indexing (similar to HNSW). Smaller upper layers stay in memory, and bottom layers with many vectors are persisted to SSTs.
ACORN addresses the problem with pre-filtering and post-filtering. Pre-filtering gives perfect recall, but it's slow with big data. We can't cut search space without losing some vectors that may match the filter. Post-filtering selects among top K vectors that ANN gives, which means the closest vectors may be filtered out. We could fetch more than K vectors, then filter. ACORN explores neighbors of neighbors (second hop) when direct neighbors are filtered out. This is slower but improves recall.
I've used a custom WAL that I've written for another side-project (it uses mem-mapped files, `rkyv`, `crc32fast`). SST files are organized in blocks (e.g 4KB per block), with compression (zstd) and Bloom filters for point lookups and range scans, which should minimize I/O amplification. Memtable size is configurable. Compaction in the LSM engine works asynchronously in the background. I do soft deletes and tombstone-based GC.
I didn't use tokio or async code in DB, just rayon and parking_lot sync primitives. The server runs with tonic and tokio. Used `rkyv` crate for serialization.
I also implemented replication (not sharding) with Raft (`raft` crate) and a gRPC server with a CLI client. Haven't properly tested raft yet. It's kind of flaky.
Here's one benchmark running on my 14-year-old laptop. Recall@5 is ~90%. About 40-50% of CPU time is spent on hashing and locking. QPS is quite low currently. For one, I can't get `simsimd` to compile using SIMD (it keeps using `simsimd_l2sq_f32_serial` for some reason), which eats ~20-30% of the CPU. I think my PC is just too old.
Will publish code if I decide to make it not horrendous.
#rustlang
English
your Ketamine-abusing, washed up, genetically-unable-to-be-cool-or-have-any-taste GOAT, everyone
Elon Musk@elonmusk
@robustus Code itself will go away in favor of just making the binary directly. The next step after that is direct, real-time pixel generation by the neural net.
English
@tsoding That misunderstanding is coming from anti-Rust people, tbf.
English
@HSVSphere "real engineering" is when you're having to deal with trash tooling instead of building reliable software.
English
@LundukeJournal It literally wasn't because of "Rust" or a memory error. It was a limit check on a data duplication that exceeded a limit.
English
September, 26: Cloudflare rewritten in “memory safe” Rust.
The change is touted as “faster and more secure” because of Rust.
blog.cloudflare.com/20-percent-int…
November, 18 (53 days later): Cloudflare has a massive outage, which took down large portions of the Internet, because of a memory error… in that Rust code.
blog.cloudflare.com/18-november-20…
English
@rustjobs_dev I like how almost every time I hear about apache, it has something to do with message brokers and stream processing.
English
🐶 Apache Iggy: Hyper-Efficient Message Streaming at Laser Speed 🦀
Iggy can process millions of messages per second with very low latency.
It is a persistent streaming log built from scratch in Rust
🔗 Link to Github repo in the comments 👇
English
@turtlefren1 @LukasHozda Rust's compiler only helps (as in, help, not remove entirely) with certain vulnerabilities.
English
Phew, thank god there wasn't a 9.3 CVE in the original sudo on 2025-06-30, a second one of 8.8 on the same day and around 20 more >5.0 ones in recent years.
CVEs only happen in Rust, guys
The Lunduke Journal@LundukeJournal
Multiple, serious security vulnerabilities found in the Rust clone of Sudo — which shipped with Ubuntu 25.10 (the most recent release). Not little vulnerabilities: We’re talking about the disclosure of passwords and total bypassing of authentication. In fact, we’re getting new reports of showstopper grade issues every few days on the Rust-based clones (like sudo, du, date, and others) which were forced to ship in Ubuntu before they were fully tested. Which is, of course, *exactly* what was predicted. But, never fear! At least these Rust clones are memory safe! PHEW!
English
@HSVSphere I don't know what Omarchy exactly is because I don't care enough to look into it, but isn't it just riced Arch? It's weird how it is this popular or how dhh is acting as if it's revolutionary. Pushing Arch to "audience new to Linux" is also funny.
English
Great position for any experienced Rust developers in need of work!
Luca Palmieri@algo_luca
Mainmatter is looking for an experienced Rust developer to join one of our projects as a freelancer. We're migrating a database from C to Rust. The project is complex and performance-sensitive. That's what makes it fun! If you're interested, send an email to the address in the screenshot.
English
I'm surprised by is how many people that I'm interviewing mentioning that they don't use coding agents due to cost for their side projects. There are in fact free agents. @AmpCode has a free tier and at the moment you can use Grok fast, still for free on OpenCode and others.
English