You're talking about this, correct?
a = b
b = a
without a "and" ?
No, no!
just about
a = b
b = c
c = 0
Am I right that a compiler could be made pretty printing a
"implicit mutual recursion spread across files!"
adding those "ands" for reading then?
Yes, exactly. Implicit mutual recursion must be regarded as error.
I'm not talking about recursive definitions at the moment.
There are advantages to the Haskell way as ell.
If you start moving functions for that reason only and start merging
you won some time when reading - but you'll have a lot more trouble when
merging..
I don't expect there being a perfect way at all.
Being forced to choose different names for similar functions
with different scopes leads to longer and harder-to-understand names.
May I conclude from that that urweb should support overloading? :)
I do not know -- there may be some issues with type checking/inferencing.
But if perfect language will be equipped with more mundane type system
it must support ad-hoc polymorphism by all means.
--------------------------------------------------------------------------------------
Do you have some specific function names in mind which share a name but
have slightly different behaviour?
No, but it is very easy to invent a dozen very realistic ones.
---- Coq ----
(* shadowing by argument (S) *)
Inductive Nat = Z | S Nat.
fold {A B S:Set} (f : A -> S -> S) (s : S) (xs : list A) : S := ...
---- Clean ----
// let-shadowing (f)
dowork f // # means "let before", as usual nonrecursive let but with silent
shadowing allowed, excellent feature!
# (f,ok_flag) = writeFile "string" f // write a string to file
| not ok_flag = f
# (f,res) = readFile 13 f // read 13 characters from a file
| isJust res = fst (writeFile ("string2 " ++ fromJust res) f)
| otherwise = f
readFile :: Int *File -> (*File, Maybe String)
writeFile :: String *File -> (*File, *
---------- Haskell ---------
{- Shadowing top level names in local defs -}
import Prelude hiding (length) -- it is not about it
length :: [a] -> Int
length xs = length xs 0 where length [] n = n ; length (_:xs) n = length xs (n
+ 1)
main = putStrLn $ show $ length "hello, world!"
would map and queryX be such cases?
Yes, perhaps.
Vag.
_______________________________________________
Ur mailing list
[email protected]
http://www.impredicative.com/cgi-bin/mailman/listinfo/ur
