Emit error when user writes incorrect format for display attribute

[magurotuna]

What's done in this PR

Resolves #127

The compiler emit an error when a user writes incorrect format for display attribute, for example:

#[derive(Display)]
#[display()] // `fmt = "..."` missing
enum Enum {}

#[derive(Display)]
#[display("struct")] // `fmt = ` missing
struct Struct;

#[derive(Display)]
#[display("custom {}", x)] // `fmt = ` missing
struct StructA {
    x: i32,
}

#[derive(Display)]
enum EnumVariantA {
    #[display(foo)] // unknown format
    A,
}

#[derive(Display)]
enum EnumVariantB {
    #[display("b", bar)] // unknown format
    B,
}

for the above examples, errors that the compiler emits look like:

error: The format for this attribute cannot be parsed. Correct format: `#[display(fmt = "...")]`
  --> tests/display.rs:62:3
   |
62 | #[display()]
   |   ^^^^^^^

error: The format for this attribute cannot be parsed. Correct format: `#[display(fmt = "...")]`
  --> tests/display.rs:66:3
   |
66 | #[display("struct")]
   |   ^^^^^^^

error: The format for this attribute cannot be parsed. Correct format: `#[display(fmt = "...")]`
  --> tests/display.rs:70:3
   |
70 | #[display("custom {}", x)]
   |   ^^^^^^^

error: The format for this attribute cannot be parsed. Correct format: `#[display(fmt = "...")]`
  --> tests/display.rs:77:7
   |
77 |     #[display(foo)]
   |       ^^^^^^^

error: The format for this attribute cannot be parsed. Correct format: `#[display(fmt = "...")]`
  --> tests/display.rs:83:7
   |
83 |     #[display("b", bar)]
   |       ^^^^^^^

Motivation

I use Display macro and I'm really happy with it. But the other day I mistakenly wrote a code like:

#[derive(Display)]
enum Foo {
    #[display("Hello Bar!")] // forgot to write `fmt = `!
    Bar,
}

I had expected it to be displayed as "Hello Bar!" and unfortunately the code review passed and it landed.
As you know, actually Foo::Bar.to_string() was evaluated as "Bar", not "Hello Bar!", since fmt = is missing.

This made me think that it would be very helpful for the compiler to emit errors for invalid format at compile-time.
I'm aware that #[display] is a valid format, but if users write parens after display like #[display()], then it most likely implies that they intend to pass fmt = "..." argument but just forget to.

Hope this patch will get landed :)

[tyranron]

@magurotuna thanks! Would you be so kind to adjust Clippy?

[magurotuna]

@tyranron Sure, done it :)

[magurotuna]

Oh I didn't realize that MSRV is 1.36.0. I'll address it.

[magurotuna]

@tyranron
Removed matches! and made some refinement around Clippy.

• set clippy's MSRV to 1.36.0 in clippy.toml and remove allow(clippy::match_like_matches_macro) attribute
• fix clippy option on CI. Currently --all flag is present, which is alias for --workspace, but it doesn't look like derive_more uses workspace. So just replaced this flag with --all-features. Also, -D clippy::all is subset of -D warnings according to clippy's readme, so I removed it.

[tyranron]

@magurotuna thanks! 🍻

[magurotuna]

@tyranron thanks for merging!
Any plan to publish the next version? I don't mean to rush you, but I would be more than happy to have the next version available, because I don't want to make mistakes of forgetting to write fmt = anymore.

[tyranron]

cc @JelteF

[JelteF]

I'll try to do release somewhere this weekend

…

On Tue, 11 May 2021, 17:08 Kai Ren, ***@***.***> wrote: cc @JelteF <https://github.com/JelteF> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#160 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAI3YJT3XWUZBCJZPNRZOFLTNFB6RANCNFSM44IKRZSQ> .

[JelteF]

In case you didn't notice I created a release last weekend: v0.99.14

[magurotuna]

Awesome! Thanks so much ❤️