I built something new this week. Wok started as a port of srtfry to Rust because, as @calle says, humans shouldn't write C. It's progressed beyond strfry now but it still shares the same super fast LMDB database under the hood. It supports more NIPs and has a handy admin dashboard for managing config and seeing basic connection stats. Feedback and contributions welcome!

Replies (17)

Based Truth's avatar
Based Truth 2 weeks ago
Rust won't save you from surveillance capitalism, Gardner. Bill Gates and pals still own your code.
eliasebner's avatar
eliasebner 2 weeks ago
Why do you say Rust is garbage? I think it has its place.
Qenis's avatar
Qenis 2 weeks ago
Keep up the great work Jeff!! My Qenis will check it out later tn image
Thanks for building it! Is it possibly deployed somewhere already? Does it support NIP-42 authentication for NIP-59 gift wrap, compatible with NIP-17 and Concord communities?
I didn't know Fil-C had progressed to something that interesting. Definitely worth having stuff like OpenSSL and OpenSSH built with Fil-C as mainstream Linux distro packages (if we will ever ensure that their complicated GC is really safe in particular), despite some performance drop. I'm personally not into coding *new* stuff in C or C++ but not due to any kind of fanaticism. I used to be a C++ dev in the past. These are pretty old languages; fewer and fewer people want to maintain such code; in a few decades these will be another Fortran and COBOL. That's okay that we switch to new languages sometimes, going through all the crises of slower compilers and whatever bad design choices (I'm personally not happy about unsafe semantics in Rust—this is too complicated; it's more reasonable to write unsafe code in C or, possibly, in Zig). If it's not Rust—I hope Rust at least influences some other new language. Also it feels more and more weird to talk about this level at all, while those who are at their peak AI psychosis stage don't care what their clankers are generating already. That could be possible, that we actually will mostly stop caring about this level, just as it happened with Assembly, but before this has happened, I still believe we'd better have good enough tools for the AI agents so they could build effectively, understand what the hell they are building, and catch more bugs at compile time.
I don't agree with the "It solves a problem nobody had", btw. I think that Fil-C in particular is not a solution for the Linux kernel, just as one example. There were attempts to build OS kernels on the best GC-based languages—all of them failed due to inadequate performance.
Nearly none. The admin dash is really only showing you connection and a few other counts. It’s very cheap.
> I missed this one. These are more general issues, unrelated to Rust specifically: putting a similar trojan into GCC (or LLVM even) would possibly be easier than into the Rust compiler; AI-generated malicious stuff is already being balanced by AI-based security checks.
I mean GC-based language *runtimes / languages with GC-based memory model, whatever. GC-based runtimes are not evil though; they give competitive performance in certain problems (message queues in particular), but they are far from something universal in that sense.
nothing built in but I'm sure an agent could write a quick script to do it for you.