I don't mean to complain. Doing software engineering for a living is a situation of extreme privilege. But there's something to be said about how alienating it can be at times. Once, just once, I want to be able to answer someone's "what are you working on? I built that".
Instead for now, I have to answer with: "well you see Not really, no. At that point, the asker is usually nervously looking at the time, and they start to blurt out some excuse to cut the conversation short, just as I was about to ask them if they'd ever heard of Conway's law.
Luckily, blogging about it is a completely different proposition. I have a bit of a reputation. You could stop reading at any time. Now would be a good time. You could stop now, too!
Still here? Alright then. There's always a lot of worry that Rust is defined by its "only implementation", rustc. But that's not true! Anyway it's still a work-in-progress and it definitely doesn't normate? But it's a thing. And then there's gccrs , which elicited some The other compiler I want to talk about is It uses rustc's lexer , but has its own parser. It has its own AST types, which leverage rowan. It uses salsa for on-demand, incremental queries like "show me all references for this symbol" , whereas rustc uses its own, pre-salsa query system.
So yeah! I didn't know that a month ago, but: rust-analyzer really does re-implement a lot of things rustc does, in a slightly different way. And we could spend time talking about rustc's librarification, but we're not going to. Instead, we'll just let those decision inform our impression of rust-analyzer as a separate project, with its own rules, its own design constraints, its own team, its own release schedule, and its own wild success β at the time of this writing, the VS Code extension alone has been downloaded almost a million times.
