Compile-time constant evaluation?

[pickfire]

Evaluate at compile-time, zero-cost and no runtime cost at the cost of longer compile time. Of course, lazy-static and similar crates would be an option.

https://github.com/rust-lang/rfcs/blob/master/text/0246-const-vs-static.md

[mgeisler]

Hi @pickfire That's an interesting idea — if it's relatively easy to implement with the existing functions?

I always assumed that the cost of wrapping strings would be very low. Also, I expect most programs will spend more time waiting for I/O to complete than waiting for strings to be wrapped.

Do you have a different use case?

[pickfire]

@mgeisler Yes, it it supposed to be very low. But then is it useful to have it it runtime when we could do it during compile time? Of course some parts still won't be able to use this feature when it involves String or ?Sized, but no reason we should not do it for &'static str which does not change. The best thing I see is, why not do it just because we can! :D

[mgeisler]

Thanks to @althonos, this is now possible. Please see his textwrap-macros crate: https://github.com/althonos/textwrap-macros. It let's you indent, dedent, and wrap strings — all at compile time.

[pickfire]

Nice, but should we mention it in README and the first page in rustdoc to make it more discoverable? Or maybe pull it into this crate behind a feature flag?

[mgeisler]

Yeah, mentioning it in the documentation is a great idea!

I don't think textwrap should have a dependency on textwrap-macros — since textwrap-macros already has a dependency back to textwrap. It seems simpler to me to have the layering work like this:

textwrap-macros
    |
    v
textwrap-macros-impl
    |
    v
textwrap

Then there is no doubt as to what version is in play at any point.