By that logic you couldn't add any new syntax or functions. Maybe they want the error about the missing symbol when calling std::tjdjtkdjr...
Really this seems extremely far fetched.
By that logic you couldn't add any new syntax or functions. Maybe they want the error about the missing symbol when calling std::tjdjtkdjr...
Really this seems extremely far fetched.
I've opened an RFC for this feature.
But there is compile_error! for that, which people should use I guess?
That XKCD remains the funniest demonstration of Hyrum's Law I've seen yet.
I wonder if we could have an easier syntax.
#[cfg((target_os = "linux" && feature = "magic") || !test)]
This is a bit more intuitive for rust users than any and all predicate.
Why any/all were used is fairly simple to note: originally, the attribute $:meta syntax was limited to well-nested meta list trees whereas today allows "list" meta to have arbitrary contents (links to the syn syntax tree enum representation).
That this doesn't look like a standard boolean test expression is also helpful, though, in nudging people to remember that cfg values can be multivalued, e.g. key = "a" and key = "b" are independent config.
This is mostly off topic to the OP topic of a blessed always-off config key though.