fix: Encapsulate Wcmd within a private wrapper

[surajk-m]

Closes #225

The change encapsulates Wcmd within a private wrapper.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 62.33%. Comparing base (29bbfa7) to head (0dd402d).
⚠️ Report is 61 commits behind head on main.

Additional details and impacted files

Files with missing lines	Coverage Δ
src/session.rs	73.53% <100.00%> (-0.05%)	⬇️

... and 3 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jonhoo]

Thanks for taking this on!

[jonhoo]

I think if you use exactly the version of issur_cmd I proposed, this impl becomes unnecessary

[surajk-m]

I was trying to use cmd: impl Into<Box<impl WebDriverCompatibleCommand + Send + 'static>> as the parameter type, but I'm getting error[E0666]

error[E0666]: nested `impl Trait` is not allowed
   --> src/session.rs:341:28
    |
341 |         cmd: impl Into<Box<impl WebDriverCompatibleCommand + Send + 'static>>,
    |              --------------^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^--
    |              |             |
    |              |             nested `impl Trait` here
    |              outer `impl Trait`

To avoid using nested impl Trait, we should flatten the types.

[jonhoo]

Ah, that just means you'll need to use "proper" generics instead:

<C, CC> where C: Into<Box<CC>>, CC: WebDriverCompatibleCommand + Send + 'static

[surajk-m]

Using generics <C, CC> with C: Into<Box<CC>> and CC: WebDriverCompatibleCommand + Send + 'static works better. The only overhead is the conversion via .into(), which is a bit more than a direct Box::new call if the type isn’t already boxed. If we only prefer non-boxed type as input, should we manually box cmd with Box::new(cmd) as Box<_> as you suggested?

[jonhoo]

I don't think Box::from is meaningfully more overhead than Box::new in this context, so let's keep it as is 👍

Could you please double-check that it's still possible pass a Box to issue_cmd though, just as a sanity check? Should just be a matter of checking that the test still compiles when you pass Box::new(GetTitle).

[jonhoo]

Just a heads up that I'll be on holiday for the next three weeks, so won't have a chance to land this until I get back :)

[surajk-m]

No worries! Wishing you a great holiday. We’ll pick this up when you're back. 😊 Safe travels!

[jonhoo]

Sorry for the delay here! I think this is now nearly ready, just two things I'd like you to test. One is the Box::new(GetTitle) from the open thread, and the other is to re-run the tool from #225 and check that webdriver indeed no longer appears in the public API. Once those checks are both good, and the conflicts resolved, then we can merge and release 🎉