Revert "Move deploy options into subcommands that actually support them" + Revert "Add R10K::CLI spec tests"

[mwaggett]

Please add all notable changes to the "Unreleased" section of the CHANGELOG.

[mwaggett]

hey @nabertrand ! We really appreciate your contribution to r10k, but this change caused the r10k CLI to no longer accept some options, notably --cachedir, which is causing issues in PE. We’re happy to work with you on any future contributions!

[nabertrand]

Hi @mwaggett! I tried to ensure that my changes didn't break anything, but apologies if I missed something. If you can provide an example usage that broke with this change, I might be able to fix it and add a spec test to make sure it doesn't break again in the future.

[nabertrand]

I opened #1048 to solve the issue I'm assuming this PR was created in response to. Let me know if there's anything else I can do.

[adrienthebo]

@mwaggett @nabertrand let's get tests passing on master, we can expand test coverage for the CLI in #1048.

[nabertrand]

@adrienthebo if master is failing with just the one R10K::CLI r10k Cri argument parsing when help specified prints command help and exits 0 test, that should be fixed in #1042, which I believe is ready for merge: #1048 (comment)

[Magisus]

So it sounds like we want to go forward with this revert for now, and re-add the tests a little differently later on, right?

[adrienthebo]

So it sounds like we want to go forward with this revert for now, and re-add the tests a little differently later on, right?

That's my vote - I'd prefer that we build the test coverage incrementally and build tests on a per-option and per-command basis for simplicity.