google
Feature idea - enumeration variant matching
Hey 👋🏾
I really like the fluid syntax of this library. While working with it, a few assertions would be great to have. Some of them are for ergonomics when working with enumeration variants. Like simple matching. So I could imagine for example something like:
enum Color {
RED,
GREEN,
BLUE,
RGB(u8, u8, u8),
}
assert_that!(Color::RED).matches(Color::RED);
assert_that!(Color::RGB(255, 0, 0)).matches(Color::RGB(_, _, _));
assert_that!(Color::RGB(255, 0, 0)).matches(Color::RGB(_, _, 0));
With the "default" API you can write these simply with assert!(matches!(...)). But then we have the syntax of the assertor it's assert_that! mixed with the other. Which does not read and feel great. So I though it might be easy enough to add this? 🙈
In any case: thanks for this library. Like it so far!
Thank you for your proposal!
Introducing pattern matching with descriptive error message is challenging as far as my knowledge goes. matches!() macro in Rust std is a syntax suger of match expression (doc), and there is no API in Rust lang to get "why it doesn't match" given a match expression and a value. So, library cannot output descriptive error message beyond assert!(matches!(...)).
It may be worth enabling users to delegate checking by passing an anonymous function; for example assert_that!(...).is(|actual_value| matches!(actual, Color::RGB(_, _, 0))). This API could be used with macros representing anonymous functions like assert_that!(...).is(matching!(Color::RGB(_,_,0))).
Interesting. Sorry, I haven't thought about the error messages. Very good point. How would this problem of descriptive error message be solved for the very generic approach you proposed? 🤔
In the generic approach in my comment, error message for assert_that!(Color::RGB(1,2,3)).is(matching!(Color::RGB(_,_,0))) can be:
Expected `matching!(Color::RGB(_,_,0))` to return true for a value `RGB(1,2,3)`, but it doesn't.
The actual value (=RGB(1,2,3)) is additional information compared to the original asserting, assert!(matches!(...)).
It may have slight difficulties in macros to stringify the delegated function though, it is possible, I think.