You haven't made it clear where exactly you're doing the imports vs @class. And I think that's causing confusion. Here's what you want to do:
- In GardenView.h, use
@class Plant
- In Plant.h, use
@class GardenView
- In GardenView.m, use
#import "Plant.h"
- In Plant.m, use
#import "GardenView.m"
This breaks the circular dependency in the headers, but still allows the implementations to see the full information of the dependent class.
Now the reason why @class alone isn't sufficient is because all @class Foo does is it tells the compiler "There exists a class named Foo", without telling it anything at all about the class. The compiler doesn't know its methods. It doesn't know its superclass. All it knows is if it sees the token Foo, then that represents a class type. You can use this in header files so you can refer to the class in arguments and return types, but in order to actually do anything with values of that type you still need the full #import. But you can put that #import in the .m file without any problem.
The only time you need #import instead of @class in your header is if you want to inherit from the class in question. You cannot inherit from forward-declared classes, so you need the full #import. Of course, you may also need the #import if you need access to other types defined in the same header (e.g. structs, enums, etc.), but that's less common in obj-c.