I'm willing to bracket mutability and make it syntactically painful if
need be. That's why I'm not too concerned about it right now.
I understand that people keep saying there is a theoretical problem
somewhere. I'm curious as to where there is a practical problem — what
specific code is problematic?
List.cons is typed as:
'a -> list('b) -> list('a|'b)
And List.first is typed as:
list('a) -> 'a
So, yes, if you add different types to the same list, and you want to
use an API beyond overlapping APIs on the types, then you're going to
have to do a match to extract the type, even if you "know" what the
right answer should be. Is that the big issue which everyone is
freaking out about?
~~ Robert.
Love Your Enemy: A Campaign to Regain Human Dignity Through Nonviolence
http://www.mettacenter.org/mc/projects/love-your-enemy
On Wed, Aug 10, 2011 at 12:24 PM, Brian Hurt <[email protected]> wrote:
> On Wed, Aug 10, 2011 at 7:29 AM, Robert Fischer <[email protected]>
> wrote:
>> Can you define "contra-" and "co-" variance for me in the context of
>> structural (not nominal) typing?
>>
>
> You get co-/contra-variance any time you have subtyping- any time you
> have a situation where all B's are also A's, but not all A's are B's.
> So if A is the type of objects having methods foo and bar, and B is
> the type of objects having foo, bar, and baz methods (with foo and bar
> having the same signatures as A's foo and bar), then B is a subtype of
> A. The fact that the subtyping relationship was introduced
> structurally or nominally is irrelevant.
>
> In terms of theory, there is no difference between structural typing
> and nominal typing. Introducing a new structural type is the
> equivalent to introducing a new uniquely named nominal interface, and
> having the interface applied to all applicable types. In practice,
> introducing that new interface and applying it to all necessary
> classes ranges in difficulty from annoying to impossible, so in
> practice there is a huge difference (and structural aka duck typing
> wins big time).
>
>> Specifically, can you show a situation where contra- or co- variance
>> are problems when using structural typing without mutable data
>> structures?
>>
>
> What are the types for List.cons and List.first? Or Map.add and Map.get?
>
> And you're going to have mutable containers, even if they're only ref
> cells. Even Haskell has MVars and TVars and the like.
>
> Brian
>
> --
> You received this message because you are subscribed to the Google Groups
> "JVM Languages" group.
> To post to this group, send email to [email protected].
> To unsubscribe from this group, send email to
> [email protected].
> For more options, visit this group at
> http://groups.google.com/group/jvm-languages?hl=en.
>
>
--
You received this message because you are subscribed to the Google Groups "JVM
Languages" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/jvm-languages?hl=en.
