Add GitHub Action declaration for installing cargo-binstall

[smallstepman]

I'd like to suggest adding action.yml, which makes it easier for users to install cargo-binstall in GitHub Action workflows, for example:

    steps:
      - uses: cargo-bins/cargo-binstall@main
      - run: cargo binstall -y ripgrep

see complete example here and the result of running that workflow.

I'm aware of https://github.com/taiki-e/install-action, but I'd prefer to:

• install directly from the source
• rely on organizations rather than individual developers
• to not have to conduct a security review of GitHub Actions that contains a few thousand lines of code and config files

❯ git clone https://github.com/taiki-e/install-action && cd install-action/
~/install-action main
❯ cloc
───────────────────────────────────────────────────────────────────────────────
Language                 Files     Lines   Blanks  Comments     Code Complexity
───────────────────────────────────────────────────────────────────────────────
JSON                        53     14136        0         0    14136          0
Shell                        6      1289       67       131     1091        232
TOML                         4        23        2         2       19          0
YAML                         4       244       15        12      217          0
Markdown                     3      1462      552         0      910          0
Plain Text                   3       201        6         0      195          0
License                      1        23        2         0       21          0
Rust                         1       773       40        33      700        108
gitignore                    1         7        1         3        3          0
───────────────────────────────────────────────────────────────────────────────
Total                       76     18158      685       181    17292        340

[NobodyXu]

P.S. I prefer "rebase against main" instead of "merging from main into this PR", but since we are doing squash merge it won't matter anyway.

If we are doing merge by commit, then the commit history would be a bit confusing.

[passcod]

lgtm generally, just a few minor bits

[NobodyXu]

CI failed due to 429 too many requests, once you update the PR I will approve and run the CI for you.

[smallstepman]

It looks like it hung again. If the error is coming from binstalk-downloader then you might want to consider using backoff crate to handle retry timings, for example:

    let retry_policy = ExponentialBackoffBuilder::new()
        .with_initial_interval(Duration::from_secs(1))
        .with_max_interval(Duration::from_secs(16))
        .with_multiplier(2.0)
        .with_max_elapsed_time(Some(Duration::from_secs(300)))
        .build();

    let response = retry(retry_policy, || async {
        match reqwest::get(&url).await {
            Ok(response) => Ok(response),
            Err(err) => Err(backoff::Error::transient(err)),
        }
    })
    .await
    .unwrap();

[NobodyXu]

It looks like it hung again. If the error is coming from binstalk-downloader then you might want to consider using backoff crate to handle retry timings, for example:

    let retry_policy = ExponentialBackoffBuilder::new()
        .with_initial_interval(Duration::from_secs(1))
        .with_max_interval(Duration::from_secs(16))
        .with_multiplier(2.0)
        .with_max_elapsed_time(Some(Duration::from_secs(300)))
        .build();

    let response = retry(retry_policy, || async {
        match reqwest::get(&url).await {
            Ok(response) => Ok(response),
            Err(err) => Err(backoff::Error::transient(err)),
        }
    })
    .await
    .unwrap();

Thanks, but we already uses backoff there, I suspect it's because we submit too many PRs tonight.

[smallstepman]

A good problem have

[NobodyXu]

Thank you @smallstepman !

[NobodyXu]

A good problem have

Hmmm, I think you have just sent an incomplete comment.

[smallstepman]

A good problem have

Hmmm, I think you have just sent an incomplete comment.

I was joking, I was referring to

I suspect it's because we submit too many PRs tonight

[NobodyXu]

Aha you mean "A good problem to have", now I get the joke

[smallstepman]

ahhh, indeed I lost a few letters ^^