Recently i've started a new project - LSP for recently released Bend 2 programming language. This is the first release which IMO is quite usable 🙂 github.com/IlyaGulya/bend2-l… Special thanks to @buildfetch for the open source build cache plan!

Turborepo 2.11 is here! → Native support for Rust, Python, and Go → Up to 4x faster startup → ZERO ISSUES LFG → devEngines.packageManager support → nub and aube support → Omit dev dependencies when pruning xitter.cf/turborepo/status/21009…

We're releasing native Remote Build Cache support for Moon! It's free for open source.

We have rewritten first production microservice at BuildFetch from Kotlin JVM to Rust… Less than 1 damn megabyte of memory for fully operating pods in Kubernetes! It has ClickHouse client, JSON & YAML serde, HTTP server we went straight with Compio (io_uring) instead of Tokio (epoll), few third-party HTTP libraries based on reqwest (Tokio), logging, metrics, basic crypto (SHA-256, HMAC, TLS) and so on. A bit of release-build optimizations, we build it via Bazel by the way: rules_rust + rules_oci, and voila! This isn't microservices anymore, it's nanoservice territory. And hey, if we had a Monolith instead of 15 separate microservices in a monorepo this would have been much much harder to pull off. And of course thank god AI models can deep research everything upfront, they code well in Rust, rustc error checking is extremely good for AI loops. The whole process took about a week of late night work in-between other product tasks, largely because we kept researching all necessary Rust libraries and other building blocks for months prior to actual migration. As technical CEO I'd say: Organizationally, just skip Go migration, no benefit in using less expressive and less type-safe language with Garbage Collection that will ruin tail latency anyway and result in bigger runtime memory footprint. Especially with coding agents. That said, there are things that drive me crazy with Rust, like this is ridiculous: - &str - &'static str - String - &mut String - &String - That's not even all! Also not the biggest fan of Rust async, coming from years of Reactive Programming. But nothing comes close to Rust in ecosystem, quality of tooling, type-system, etc for a native no-GC runtime. Zig is just too risky right now, current leadership is questionable and we don't have time to re-create all basic libraries. Maybe in few years, we will keep an eye on it for sure.

Introducing CUDA Rust! CUDA Rust lets you write GPU kernels natively in Rust, not just launch them from it. Two paths: cuda-oxide for SIMT kernels compiled to PTX, and cutile-rs for Tile-based programming on stable Rust. Both can catch aliasing errors at compile time. Technical blog: nvda.ws/4hm1bHS

We are releasing Buck2 Remote Cache support today 🦌

Small detail about BuildFetch project dashboard graphs: we render graphs/bars with relative transparency When you have 550 requests of type A in the same bucket as 6k requests of type B, type B matters more We calculate relative transparency and render all data points at the same time with visibly different importance This applies to Cache Reads / Write smooth graphs as well as Cache Hit Ratio bars Small details matter!

We've rolled out native Nx Remote Cache support BuildFetch Cache now provides extremely fast build cache for TypeScript and JavaScript frontend - backend Monorepos, React Native and Electron desktop apps. Free for Open Source 🫡

Sccache Remote Storage is now natively supported by @BuildFetch Cache allowing long compilation tasks to be cached and reused across machines. Free for Open Source! Sccache includes support for caching the compilation of @rustlang, Assembler, C/C++, NVIDIA CUDA through nvcc and clang, and AMD ROCm HIP for AI and ML. There are extremely complicated projects like @zeddotdev (written in Rust), as well as AI labs with @NVIDIAAI and @AMD native code that will benefit from native Sccache support in addition to our native support of @bazelbuild build system. BuildFetch is moving quickly towards being a unified Cache system for the full stack: Backend, ML / AI, Frontend & Mobile. Both in Cloud and On-Prem.