• Post Reply
  • Bookmark Topic Watch Topic
  • New Topic

Generics

 
Tony Morris
Ranch Hand
Posts: 1608
  • 0
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I can create a type as follows...



...then I can use it...



...but how can I enforce that the returned type is of exactly the parameterized type, and not a subtype? i.e. In the above example, how can I enforce that an object of type java.lang.Object is returned, and not just any subtype of java.lang.Object (such as java.lang.String)?
 
Stefan Wagner
Ranch Hand
Posts: 1923
Linux Postgres Database Scala
  • 0
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
I guess you can't, because if you have to return an object of type RT, every subtype SRT of RT is allowed.
An SRT is an RT.

You may only define an own return type and declare it final.


right?
[ January 17, 2005: Message edited by: Stefan Wagner ]
 
Tony Morris
Ranch Hand
Posts: 1608
  • 0
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Sure you can do that.
Suppose you don't know the type in advance.
Specifically, how could you write interface I such that that parameterized type T is exactly the type that is returned by implementations, enforced at cimpile-time?
 
Jim Yingst
Wanderer
Sheriff
Posts: 18671
  • 0
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
[Tony]: Sure you can do that.

I don't see how. I don't believe this is possible. Why would it be necessary? Why would it be a problem if an implementing class returned Strings rather than Objects?
 
Tony Morris
Ranch Hand
Posts: 1608
  • 0
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
"Sure you can do that" - I was responding to the post that mentions a class being declared final implies that if it is used as the parameterized type, then you are guaranteed that no subclass types will be returned.

There are many reasons why you might want to guarantee the type of the object, returned by a method (rather than just being guaranteed that, whatever it is, it is assignable to the reference type returned) - the difference in this case, is that the type is unforeseen and so a compile-time enforcement is not so straight-forward.

I had a quick play around with composites of wildcards to no avail - surely it must be possible.
 
I agree. Here's the link: http://aspose.com/file-tools
  • Post Reply
  • Bookmark Topic Watch Topic
  • New Topic