*: Migrate to Rust 2021 edition

[nuke-web3]

Confirmed to build locally on Ubuntu 20.04.3 LTS:

rustc 1.56.1 (59eed8a2a 2021-11-01)
rustc 1.58.0-nightly (e90c5fbbc 2021-11-12)

[mxinden]

Thanks @nukemandan for the upgrade work!

The most important rule for editions is that crates in one edition can interoperate seamlessly with crates compiled in other editions. This ensures that the decision to migrate to a newer edition is a "private one" that the crate can make without affecting others.

https://blog.rust-lang.org/2021/05/11/edition-2021.html

If I am not mistaken there is no downside for rust-libp2p to upgrade to edition 2021. Anyone objecting?

Not sure what the best practice is, but I would still consider an upgrade to 2021 to be a breaking change. Thus this pull request would need to add a changelog entry to each crate's CHANGELOG.md and bump the version if not already done (look for [unreleased] in CHANGELOG.md). See release docs for details.

[thomaseizinger]

If I am not mistaken there is no downside for rust-libp2p to upgrade to edition 2021. Anyone objecting?

AFAIK it will bump the MSRV to 1.56. Are we okay with that?

[nuke-web3]

It is also strongly suggested that we included a minimal rust version to use, specifically 1.56.1

Cargo.toml should include:

rust-version = "1.56.1"

https://doc.rust-lang.org/cargo/reference/manifest.html#the-rust-version-field

it cannot include semver operators or pre-release identifiers.
The rust-version must be equal to or newer than the version that first introduced the configured edition.
The rust-version may be ignored using the --ignore-rust-version option.

More discussion:

• move to rust 2021 paritytech/parity-wasm#322 (comment)
• Check MSRV only with minimal-version dependencies Tamschi/rust-template#28

[mxinden]

AFAIK it will bump the MSRV to 1.56. Are we okay with that?

I am fine with it. Anyone out there objecting?

It is also strongly suggested that we included a minimal rust version to use, specifically 1.56.1

What is the benefit of explicitly setting the version? Isn't it already implicitly set by moving to Rust 2021 @nukemandan?

[nuke-web3]

I have not played with behavior for different versions with and without this included, but I believe it enforces builds with no less than this version, and in this case the 1.56.1 (with the security patch) will be required to build, without specifying a flag to ignore it.
This will theoretically avoid issues that using older versions may introduce. But to be honest, it's not clear (I have not done a lot of research) if this is worth enforcing. My intuition is to add it.

[nuke-web3]

rust-version = "1.56.1" introduced

For reference, I tried to get this to cargo build with 1.56.0 and it failed with

error: package `...` cannot be built because it requires rustc 1.56.1 or newer, while the currently active rustc version is 1.56.0

With --ignore-rust-version it does build with 1.56.0, as expected.

Should be good to go with a squash merge to make the commit log clean on master IMHO 👍

[mxinden]

rust-version = "1.56.1" introduced

For reference, I tried to get this to cargo build with 1.56.0 and it failed with

error: package `...` cannot be built because it requires rustc 1.56.1 or newer, while the currently active rustc version is 1.56.0

With --ignore-rust-version it does build with 1.56.0, as expected.

Should be good to go with a squash merge to make the commit log clean on master IMHO +1

Thanks for testing. I understand that rust-version enforces usage of > VERSION in downstream dependencies, though I wonder whether we need to set it explicitly, or whether the implicit version restriction through edition = "2021" isn't enough?

[thomaseizinger]

Thanks for testing. I understand that rust-version enforces usage of > VERSION in downstream dependencies, though I wonder whether we need to set it explicitly, or whether the implicit version restriction through edition = "2021" isn't enough?

I think it should give a much better error message? At least for version 1.54 on-wards (where rust-version stabilized). Not a huge benefit I guess?

[mxinden]

Thanks @nukemandan for the help!