From: Xavier Noria Date: 2010-07-08T20:40:35+09:00 Subject: Re: Complie-time polymorphsim and Run-time polymorpshism On Thu, Jul 8, 2010 at 12:55 PM, hari bhat wrote: > Hi > >                    I wanted to know about complie-time polymorphsim > and run-time polymorpshism in ruby As a rule of thumb Ruby is runtime-only. When you evaluate obj.method obj's class is checked at that exact point in the execution of the program. If you later execute that line again, it is checked again. And due to the nature of Ruby obj could be an instance of a different class, or of the same class but with different API than before. You check it always. There's a gotcha with core classes though, like Hash and friends. If you for example subclass Hash, MyHash, and define a custom #[]= that's not seen by the rest of inherited methods. So for example an inherited MyHash#merge! is still going to use the original Hash#[]=. Reason is Hash is implemented in C and #merge! has hard-coded calls to the C function that implements Hash#[]=. So polymorphism is missing here. Same with String and others. It feels a bit strange in an otherwise pure OO language like Ruby, but as far as I know it is a pragmatic compromise for speed. Rubinius and JRuby would naturally support that polymorphism, but they don't on purpose for compatibility with MRI.