Graceful component registrar

[joshdifabio]

Some workflows can result in ComponentRegistrar::register() being called multiple times for the same component. In order to avoid messy, repeated boilerplate in client code, it makes sense for the register() method to be null-op in cases where the component has already been registered with the same path.

Related to #3871

[vkublytskyi]

If some workflow register same component several time there is an issue in workflow. So the reason why same component (from same path) registered more than once should be fixed.

Initial PR #3871 tries to solve other issue when Magento installation have several copies of same component in different folders (e.g. in app/code and vendor)

[vkublytskyi]

In any case thank you for your impact on this issue. Let's collaborate and find optimal solution

[joshdifabio]

Initial PR #3871 tries to solve other issue when Magento installation have several copies of same component in different folders (e.g. in app/code and vendor)

I see. In that case, as you said, this PR isn't relevant to that issue. I'll stay out of that one as I don't have any experience of such a scenario.

Putting that issue aside, I still see this PR as an improvement; what is the value in throwing an exception in the case that a component is registered twice with the same directory path? Will we start seeing developers wrap their register() calls in a try-catch and checking the existing path in the catch block? Does it not make more sense for register() to be idempotent?

[mazhalai]

@joshdifabio please sync with the develop branch and rerun travis builds.

[joshdifabio]

If at any point it's decided to merge this then I'll be happy to do so, but it's a waste of my time otherwise.

[buskamuza]

@joshdifabio , could you, please, provide an example of the workflow that leads to registration of the same component multiple times?

[joshdifabio]

@joshdifabio , could you, please, provide an example of the workflow that leads to registration of the same component multiple times?

Not off the top of my head, but this is more about throwing an exception in a situation where doing so provides zero benefit. However, it's a complete waste of time (which is extremely precious to many people, myself included) and completely missing the benefits of OSS to spend so much time discussing such minor changes. This PR is absolutely not worth the time being spent on it, so I'm going to close it.