|
|
|
|
|
by MaxBarraclough
337 days ago
|
|
Sounds like progress is being made on the language spec front, that's good to see. I'm not following what you meant by this though, it seems like a contradiction: > A language specification is not required to be qualified. The behavior of the compiler needs to be described. But they're putting work into reviving the language spec, to enable certification? Also, if the source language hasn't been described, then surely the compiler's behaviour hasn't been described. Or did you mean that their documentation is for the Ferrous flavour of Rust and might not reflect the latest version of the Rust language? |
|
It has already been qualified. Upstream has always wanted a spec. It’s being worked on because it’s desirable, not because it’s blocking safety critical cases.
You’re always going to need to have more than a language spec because you qualify compilers not languages.
> Also, if the source language hasn't been described, then surely the compiler's behaviour hasn't been described.
It has. At least to the degree that regulators find it acceptable.
> Or did you mean that their documentation is for the Ferrous flavour of Rust and might not reflect the latest version of the Rust language?
There is no difference in flavors, but it is true that each version of the spec is for a specific version of the compiler, and so sometimes that will lag a bit. But that’s just how this process works.