Don't clone when with_fallback is called

[pocke]

Currently, with_fallback behavior is complexity. This cause is cloning in with_fallback method.
This change is required from #13 .

[k0kubun]

Tests are broken. Do you intend to change that behavior?

[pocke]

Yes.

Currently behavior

c = Command.new do
  xxx
end

c_fallback = c.with_fallback do
  xxx
end

c == c_fallback # => false

Changed behavior

c = Command.new do
  xxx
end

c_fallback = c.with_fallback do
  xxx
end

c == c_fallback # => true

So, the following test is broken. https://github.com/cookpad/expeditor/blob/master/spec/expeditor/command_functions_spec.rb#L179-L192
Because it depends the difference c and c_fallback.

However, in real, I think the case doesn't exist.

[k0kubun]

So with_fallback is changed to be destructive, right? In my opinion, with_fallback is reasonable to be non-destructive. If the method's behavior is changed to be destructive, I think its name should be changed to set_fallback or something.

[pocke]

So with_fallback is changed to be destructive, right?

yes

I think its name should be changed to set_fallback or something.

👍
I'll deprecate wtih_fallback and add set_fallback. The behavior between the two methods are same. Is it ok?

[k0kubun]

The behavior between the two methods are same. Is it ok?

It means that it breaks backward compatibility. If you update version properly (0.5.0, since it's < 1.0), I think it'd be acceptable.

[pocke]

I've deprecated the method. And I've added an entry of this PR into CHANGELOG.md .
I think we can release the change with other changes( #13 and any refactor ).

[k0kubun]

LGTM

[adorechic]

LGTM