Increase syn dependency lower bound to 2.0.0

[mulkieran]

Closes #3650

[mulkieran]

stratisd_proc_macros compiles...but panics.

[mulkieran]

Now panicking in a better place...

[mulkieran]

compiles

[mulkieran]

If it passes, it should be good.

[jbaublitz]

I think this can be further destructured using this enum. It will allow you to avoid converting to a String and then trimming the quotation marks, instead just getting the String value.

[mulkieran]

litstr type is TokenTree::Literal, which is a struct with private fields and a rather awkward set of methods in syn2. I think that syn2 probably expects that the normal way to encode that argument value is without the '"' in the first place, as a plain Ident, but I don't want to hit every spot where the pool_mutating_action annotation is used to change that, so I left it alone.

[jbaublitz]

Ah, I see. I think changing this to use an ident would be better after looking at the types. Would you be willing to do that? I think it would generally be better from the standpoint of upgrading to the new API. I'll also look into the recommended way of doing this because I have some other crates I maintain that need updates.

[mulkieran]

I would, but I'ld like to postpone that until we start the next release. I'll file an issue.

[jbaublitz]

Okay sounds good!

[mulkieran]

@jbaublitz Ok, it is ready again.

[jbaublitz]

I'm retracting my approval just to suggest one more possible avenue. Does parse_nested_meta do what we want here? Maybe not. I'll look into it more when I get back, but it looks promising.

[mulkieran]

@jbaublitz The last commit tightens it up a bit. It can be tightened up even further by specifying parse_args with Ident type, but that requires changing all the uses every where to actually be Idents, not strings.

[jbaublitz]

Would you mind holding this until I get back? I want to investigate further.

[mulkieran]

@jbaublitz Yeah, that's fine.

[jbaublitz]

@mulkieran I found a better way to do it:

            if let Ok(level) = a.parse_args::<LitStr>() {
                let err_str = level.value();
                (i, level.parse::<Ident>().expect(format!("Could not parse value {err_str} for rejection level as an Ident").as_str()))
            } else {
                panic!("pool_mutating_action attribute must be in form #[pool_mutating_action(\"REJECTION LEVEL\")]")
            }

This requires no changes to the way we currently provide the rejection level.

[packit-as-a-service]

Congratulations! One of the builds has completed. 🍾

You can install the built RPMs by following these steps:

• sudo yum install -y dnf-plugins-core on RHEL 8
• sudo dnf install -y dnf-plugins-core on Fedora
• dnf copr enable packit/stratis-storage-stratisd-3682
• And now you can install the packages.

Please note that the RPMs should be used only in a testing environment.

[mulkieran]

@mulkieran I found a better way to do it:

            if let Ok(level) = a.parse_args::<LitStr>() {
                let err_str = level.value();
                (i, level.parse::<Ident>().expect(format!("Could not parse value {err_str} for rejection level as an Ident").as_str()))
            } else {
                panic!("pool_mutating_action attribute must be in form #[pool_mutating_action(\"REJECTION LEVEL\")]")
            }

This requires no changes to the way we currently provide the rejection level.

Yeah. That's a better way to parse. But the expect is still rejected, so I'll stick with panic.

[jbaublitz]

What's rejecting it?

[mulkieran]

It's a correct clippy lint: https://rust-lang.github.io/rust-clippy/master/index.html#/expect_fun_call . With an expect, the error message is eagerly evaluated. With the panic, it is not.

[mulkieran]

squashed.