Ben Schulz wrote: [...] > Section 4.1 of the JLS disagrees with you, the null type is there, you > just can't name it as part of a Java sentence. And it's a pity that > it's so, but I understand the reasoning behind it ( > http://blogs.sun.com/abuckley/entry/naming_the_null_type ). >
As a sidenote: while I think a type system should be a lattice, the bottom type by itself is not sufficient (as stated in that blog post). The reason is that a lattice requires you to have exactly one largest common subtype for each type (and the dual), which can easily be broken by multiple inheritance: interface X extends A,B interface Y extends A,B Assuming no other types in between, we have both X and Y as _a_ largest common subtype, but the uniqueness constraint is not fulfilled. It doesn't matter if there is a bottom or not, in fact the bottom element is usually derived as a feature of a complete lattice, not there by definition. Having an explicit join would fix it, since (A & B) would be a subtype of A and B as well as a supertype of both X and Y and thus we got a Dedekind-McNeil completion into a lattice (X | Y would be the same in a closed world). If you don't have that defining the type of e.g. a parameter of a method that expects both A and B as interfaces would not be possible in a safe way. With generics you can actually do that in Java, but not without. Does that help fixing the mess with nulls? Probably not :-) I just like lattice theory as a conceptual modeling tool. Peter --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "The Java Posse" group. To post to this group, send email to javaposse@googlegroups.com To unsubscribe from this group, send email to javaposse+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/javaposse?hl=en -~----------~----~----~----~------~----~------~--~---
