[Prev][Next][Index][Thread]

Re: more than covariance, distributivity?



"Benjamin C. Pierce" <bcpierce@saul.cis.upenn.edu> writes:

> Before forwarding this to the list, let me ask whether you're aware of
> the work that's been done on intersection and union types? 

yes

> These kinds
> of distributive laws come up all over the place (for distributing unions
> and intersections over arrow types, over each other, through records,
> etc.)

Yes I know. Most subtyping I know use the distributivity in any case,
making approximations for the sake of simplicity.
(eg: [ [ `A ; `B ] ]  and  [ [`A] ; [`B] ]  having the same type)
What i'm trying to see is in which case these are not approximations, but
really equality.

But maybe i'm explaining badly, or this really is known/trivial and i missed
the evidence.

> Starting points for reading about this stuff are
> 
>    - work on union types in XML-processing languages.  For example, see
>         http://www.cis.upenn.edu/~bcpierce/papers/regsub.ps

I've not read this one yet. I did try to read as much as i can in the others.
I'll try to see if my problem appears in XDuce.

Sorry to bother you.

--
Pixel