Clap v3 update: keygen

[KirillLykov]

Problem

Update clap to version 3. This PR is part of the effort to migrate the whole repo to clap-v3

Summary of Changes

• update CLI to use clap v3
• switch from clap-utils to clap-v3-utils

[codecov]

Codecov Report

Merging #24479 (f15dbe8) into master (3155270) will increase coverage by 11.8%.
The diff coverage is 67.1%.

❗ Current head f15dbe8 differs from pull request most recent head cd87e5d. Consider uploading reports for the commit cd87e5d to get more accurate results

@@             Coverage Diff             @@
##           master   #24479       +/-   ##
===========================================
+ Coverage    70.0%    81.9%    +11.8%     
===========================================
  Files          36      592      +556     
  Lines        2255   163645   +161390     
  Branches      322        0      -322     
===========================================
+ Hits         1580   134102   +132522     
- Misses        560    29543    +28983     
+ Partials      115        0      -115     

[t-nelson]

doesn't look like grind_validator_starts_with strictly requires a String. I'd prefer to change its signature to take &str

[KirillLykov]

thanks for pointing out, fixed all the occurrences except for one which is more elaborated

[t-nelson]

ditto

[t-nelson]

what's constraining the v3 version of is_parsable to taking a String?

[t-nelson]

superfluous closure?

[KirillLykov]

without this closure it doesn't compile with error:

error: implementation of `FnOnce` is not general enough
   --> keygen/src/keygen.rs:500:26
    |
500 |                         .validator(is_prompt_signer_source)
    |                          ^^^^^^^^^ implementation of `FnOnce` is not general enough
    |
    = note: `fn(&'2 str) -> Result<(), std::string::String> {is_prompt_signer_source::<&'2 str>}` must implement `FnOnce<(&'1 str,)>`, for any lifetime `'1`...
    = note: ...but it actually implements `FnOnce<(&'2 str,)>`, for some specific lifetime `'2`

error: could not compile `solana-keygen` due to previous error

I found explanation here: "The issue stems from how closures, type inference, and higher-order signatures play together: for historical reasons, closures favor type inference / are reluctant to get higher-order promoted. In fact, they "rarely" do, except for one special hard-coded case ... "
Looks like it's hard to do something with it

[t-nelson]

even when adapting the signature of is_prompt_signer_source to explicitly take &str? I'd think it's just not able to resolve AsRef<str> to &str automatically 🤔

[KirillLykov]

If I use &str it works (updated). Looks like validator is always assuming &str so might be unnecessary to have generics there. My only concern is that all the functions in the input_validators.rs are using the same generic signature and may be there is a reason behind which is unknown to me. So I got rid of generic only in is_prompt_signer_source, others I left as they are to modify in next PRs by demand, if makes sense.

[t-nelson]

cool! agreed updating all of the validators in a future PR is the right thing

[t-nelson]

lgtm. thanks!