dtolnay/efg

Conditional compilation using boolean expression syntax, rather than any(), all(), not()

efg's Issues

specifying ranges

rust-lang/rfcs#3036 describes a family of predicates to target minimum versions of platform APIs. For example min_libc_version would return the version of libc. It was expected that max_* variants would be specified eventually as well ¹, allowing ranges of OS versions to be targeted.

Unfortunately the RFC hasn't seen any updates in a while, but the challenges it attempts to address still exist. I'm wondering whether efg could help prototype syntax to specify ranges within attributes. It'd be neat if we could target a specific version range Windows by writing something like this:

#[cfg(windows && os_version >= "6.0.6000" && os_version < "7.3.5000")]

Or possibly taking it further and allowing shorthands such as:

#[cfg(windows && "6.0.6000" <= os_version < "7.3.5000")]

I'm not sure what this feature should look like, but I think having the ability to specify ranges using shorthands would be neat as well. What do you think?

Footnotes

1. Being able to specify the maximum OS version is useful when you're writing polyfills for older versions. This would allow a developers to say something like: "This code runs on windows 10+ only", which can be leveraged to produce more optimized code. ↩

Suggestion: change `=` operator to `==` and add `!=` in 0.2

First of all, I think this crate is a great idea! It always irked me how verbose the syntax for cfg was compared to regular Rust.

Suggestion: if we're going for syntax simplification, I'd love to also simplify frequent not(feature = "somefeature") not only to !(feature = "somefeature") but all the way to feature != "somefeature".

It feels a lot more natural to use != operator, but I suspect it will also require changing = to more Rust-like ==. It would be a breaking change, but perhaps a good thing for consistency with the rest of Rust syntax.