#3583: Default view patterns
-----------------------------------------+----------------------------------
Reporter: ksf | Owner:
Type: feature request | Status: new
Priority: normal | Milestone: _|_
Component: Compiler | Version: 6.10.4
Resolution: | Keywords:
Difficulty: Moderate (less than a day) | Os: Unknown/Multiple
Testcase: | Architecture: Unknown/Multiple
Failure: None/Unknown |
-----------------------------------------+----------------------------------
Comment (by ksf):
Well, I think that totally implicit view patterns are probably a very good
idea, because they reduce the syntactic overhead of abstraction (I'm going
to pretend to be able to override (:) for the sake of argument):
{{{
foo (x:xs) = x + length xs
vs.
foo (-> (x:xs)) = x + length' xs
vs.
foo xs = head' xs + length' (tail' xs)
}}}
Consider parsing libraries: Only uu-parsinglib and Parsec (>=3) support
matching on arbitrary left-disectable sequences, while certainly every
parsing library could do that. I think the main reason for this is the
clarity and ease of matching on (:). While # 2 does certainly look more
idiomatic than # 3, I fear it would still not be enough to foster wide-
spread adoption of such abstraction. Even now, uu-parsinglib and Parsec3
come with separate uncons classes. The same argument applies to [Char] vs.
ByteString vs. Data.Text vs. Seq Char: gazillions of functions could work
on all of them, and most likely being made to work (with a bit of luck)
just by adding an -X flag (I'm in the "safe Haskell by throwing
typeclasses at it"-camp).
As I can't see any instance where # 1 and # 2 would differ in anything but
how often the method "view" can be captured, I certainly prefer the # 1
for its clarity:
Providing two language options -- e.g. -XViewPatters and -XImplicitViews
-- might not only help avoiding Wadler's Law and offload the discussion to
what's common usage after some time, but also provide a graceful update
path for code that already has "view" in scope: -XViewPatters would only
capture it with explitit -> Syntax, -XImplicitViews always. In any case,
if things won't work out, you'll get a type error.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/3583#comment:9>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
