On 04/09/2012, at 1:39 PM, john skaller wrote:
>
> On 04/09/2012, at 11:38 AM, john skaller wrote:
>
>> Last night i said to Shayne: "Felix needs more magic"
>
> ftp://ftp.cpsc.ucalgary.ca/pub/projects/charity/literature/manuals/manual.ps
So for a union like
union X [A,B] = | X1 of A | X2 of A * B | X3 of A+B ) ...
map is clearly:
fun map[A,B, A', B']
(fa: A -> A')
(fb: B -> B')
(x:X[A,B])
: X[A',B'] = {
match x with
| X1 ?a => X1 (fa a)
| X2 (?a,?b) => X2 (fa a, fb b)
| X3 (?k) => match k with
| case 0 ?a => case 0 (fa a)
| case 1 ?b => case 1 (fb b)
In more compact form, A * B can be handled itself by a (trivial) map,
as can A+B. If we run into an X[A,B] valued subterm, we just reapply
the map.
This machinery cannot handle exponentials, i.e. functions, i.e. terms
of type A -> B etc. Given some k: A -> B we can compose it:
inmap . k . outmap
where inmap: A' -> A, outmap B -> B'
We can take outmap as fb, but inmap would have to be the inverse of fa.
In general then, we have to provide, for each type variable, a pair of functions
which are inverses. Note this also applies to arrays (since they're
exponentials).
If run into a term of type X[ h (A,B), j (A,B) ] where h or j is not the first
or second projection, we have polymorphic recursion, and we have to
construct new fa, fb maps for h and j. Ouch!
Fold is similar. For one type variable it's trivial,
it's just a simple visitor.
--
john skaller
skal...@users.sourceforge.net
http://felix-lang.org
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Felix-language mailing list
Felix-language@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/felix-language
