PVF: refactor errors to use thiserror

[mrcnski]

We're already using thiserror in some places, so we should consider expanding its usage to reduce boilerplate. For example, we could add #[from] attributes for the variants of ValidationError, making it easier to construct. For sure another candidate is PrepareError, we could provide an auto conversion for e.g. IoErr, reducing the boilerplate whenever it's constructed. Then instead of a bunch of map_errs on std::io::Error, we can just use ?.

See here for an example thiserror usage. cc @eagr since you were interested in some refactorings. :)

[maksimryndin]

Hello @mrcnski ! Can I take this issue as the first one to familiarize myself with the codebase?

[mrcnski]

@maksimryndin Hey 👋🏻 That would be great! Please let me know if you have any questions.

If it helps, we have a README in the pvf/ crate, some module docs (run cargo doc --open --document-private-items to make sure you see them all), and some docs in https://github.com/paritytech/polkadot-sdk/tree/master/docs/contributor as well.

[maksimryndin]

Thank you @mrcnski So a roadmap for now

• check https://github.com/paritytech/polkadot-sdk/tree/master/docs/contributor
• read pvf/ README and some pages of the book
• fix broken doc links
• fix codec macro typo https://github.com/paritytech/polkadot-sdk/blob/master/polkadot/node/core/pvf/common/src/error.rs#L81 (see the comment below)
• refactor ValidationError and PrepareError to use thiserror
• run unit tests
• setup Linux amd64 working environment
• run integration tests
• run zombienet tests

[maksimryndin]

@mrcnski codec macro was mistakenly applied two times to Kernel error (so it was encoded with 10 instead of 11 and the same as JobDied). I changed it to 11 because

1. it was an initial goal of the code author
2. Kernel is less frequent than JobDied so in case of existing error encoding it is more probable to have 10 as JobDied than Kernel

Output of cargo expand

// before

PrepareError::JobDied { ref err, ref job_pid } => {
    __codec_dest_edqy.push_byte(10usize as ::core::primitive::u8);
    ::parity_scale_codec::Encode::encode_to(err, __codec_dest_edqy);
    ::parity_scale_codec::Encode::encode_to(
        job_pid,
        __codec_dest_edqy,
    );
}
PrepareError::Kernel(ref aa) => {
    __codec_dest_edqy.push_byte(10u8 as ::core::primitive::u8);
    ::parity_scale_codec::Encode::encode_to(aa, __codec_dest_edqy);
}

// after

PrepareError::JobDied { ref err, ref job_pid } => {
    __codec_dest_edqy.push_byte(10u8 as ::core::primitive::u8);
    ::parity_scale_codec::Encode::encode_to(err, __codec_dest_edqy);
    ::parity_scale_codec::Encode::encode_to(
        job_pid,
        __codec_dest_edqy,
    );
}
PrepareError::Kernel(ref aa) => {
    __codec_dest_edqy.push_byte(11u8 as ::core::primitive::u8);
    ::parity_scale_codec::Encode::encode_to(aa, __codec_dest_edqy);
}

[maksimryndin]

@mrcnski I am not sure on labels as I think it is also worth to add R0-silent and T11-documentation (although I've checked myself the fixed links)

PR here #2958

I need some time to prepare a proper working environment as it requires Linux amd64 and more free disk space in order to run integration tests

[mrcnski]

I need some time to prepare a proper working environment as it requires Linux amd64 and more free disk space in order to run integration tests

Yes, unfortunately getting this environment set up and testing on it can be a pain. I rented a VPS, but it is really slow to compile.

I think you don't need it for this particular PR though. Most of the Linux-amd64-only tests are testing things like spawning processes, security features, etc. And if there is a failing test, CI will catch it for you.

[maksimryndin]

Thank you for a warm welcome @mrcnski :)

[mrcnski]

No problem @maksimryndin, let me know if you have any questions. And I appreciate the thoroughness!