https://xitter.cf/t.co/c7VTlLqKH2
Joined July 2011
- Tweets4.8K
- Following581
- Followers1K
- Likes29.3K
David Barsky retweeted
very excited to see this out of A\
this is our thesis at @erscio ! But we are building a git compatibility layer: ersc.io/blog/what-comes-afte…
enterprise adoption of new source control is a big ask, so making a smooth bridge from the past to the future is a big part of our offer
New on our Engineering blog: Creating in-house source control.
In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale.
Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
holy moly @superlogical's rex is *so darn good*. just oozes with polish, style, and care.
a tip for friends new to Rust: Geekbench's single-core CPU benchmarks are an unreasonably helpful proxy for how snappy Rust compilation feels on a warm development loop.
i mean, miri isn’t really community-enforced scrutiny, it’s a project-endorsed tool that provides mechanisms to verify whether unsafe code is sound. it’s certainly more advanced than anything i’m aware of that touches the JNI!
and i’m curious: how would you feel if unsafe was a nightly-only/stdlib-only feature? because that’s not too dissimilar to the situation you’ve got on the JVM.
agree to disagree, but to move the conversation a few steps back: you can’t simply remove sun.misc.unsafe and expect your program to work. you still have the JNI and the runtime itself. how is the runtime speaking to the os?
and yes, a runtime like the JVM or WASM help reduce the impact of soundness issues in Java or Scala, but a lot of places are using Rust where a runtime can't work (I include a lot of latency-sensitive developer tools in this!) or the alternative is *definitely* unsound.
rust does have a complicated type system—no argument from me—but nobody's found a fundamental soundness flaw within the overall design. people tried! i know that's not the same thing as a positive proof, but given it's been ~10 years, i know how I'm willing to bet.
put differently, there are a lot of mechanisms to surface unsound code + the community has developed extremely strong norms against using unsafe code. if unsafe code is used, it's generally heavily quarantined.
sorry, i wrote too many exclamation marks. what i'm trying to say: it's easier to audit for unsafe and me running miri during development/CI surfaces potential issues in the libraries that i'm depending on. a potential issue was surfaced for me yesterday!
like, a big part of what was problematic about the actix-web situation was the larger community's reaction, but part of that reaction was the development of norms and tools to avoid unnecessary usage of unsafe! a lot of libraries now have miri running in CI!
there might be confusion between and @tigregalis; they never said "usage is pretty judicious".
my source of data is being in the review loop for people importing libraries into facebook's monorepo and facebook + maintaining some pretty popular libraries in the rust ecosystem.
anyways, i got annoyed by the wikipedia article and proposed a bunch of edits: en.wikipedia.org/wiki/Talk:M…
as for that CVE crate: it's not really used and it's a soundness hole that'll be fixed: github.com/rust-lang/rust/is…. nobody's writing that code in the wild.
as for unsafe: i think usage is pretty judicious and the data i've seen tells that it's not a cause for concern.
anyways, i get it if you're done with this conversation—I would be too!—but i'd be very careful about expanding the definition of memory safety to include "does not leak memory" because that'd mean java et. al. are not unsafe!
i don't think that the wikipedia article is correct, partly because the ISRG (memorysafety.org/docs/memory…) disagrees, partly because because the CWE referenced by the article points to a deprecated definition (that has been superceded by cwe.mitre.org/data/definitio….
