Cleanup GraphQL Localization

[MikeAlhayek]

No description provided.

[gvkries]

I'm not sure if we should do this, after all the discussions about breaking changes lately. The where types can be used directly, even if it is not the case in OC itself. Also, I don't see anything that could be made abstract in the base classes. So I don't see enough advantage of introducing another breaking change.

Anyway, if we do it, the c'tors should be protected.

P.S.: The PR description is misleading, why is it related to localization?

[MikeAlhayek]

Anyone project is able to inherit these classes. Now they have to provide the IStringLocalizer into the base constructor. The base constructor should accept any IStringLocalizer not only IStringLocalizer<>. This way the derived classes don't need to resolve the wrong Localizer to pass to the base class.

It's abstract because these classes don't exist alone. They are base classes. Aren't they? Since I replaced IStringLocalizer<> with IStringLocalizer, this will cause an exception if someone added it to the IoC and tries to resolve it.

[gvkries]

Oh, I added that parameter, I forgot about it. Than it's already a breaking change introduced by myself. We could either fix that or you can go ahead, I'm fine with whatever you want.

[MikeAlhayek]

We could either fix that or you can go ahead

That is what this PR fixes. Do you see something else that I missed?

[gvkries]

We could either fix that or you can go ahead

That is what this PR fixes. Do you see something else that I missed?

My change is still a breaking one after this PR, because the parameter less c'tor is missing. I mean we can either make the string localizer optional with a default one so my change is not breaking anymore, or go ahead and formalize the actual intent of these classes being base classes. That's how I understand your change. Isn't that correct?