[Haskell-cafe] Re: Overloading functions based on arguments?

[Luke Palmer]

On Sat, Feb 14, 2009 at 7:56 AM, John A. De Goes <john at n-brain.net> wrote:

> Don't overlook the advantages of using familiar operators and names: you
> have some intuition about '+' and 'map', so if you see them, then you'll
> have some idea what they do (assuming the author is neither stupid nor
> malicious). However, if you see some operator like '$>+' or some name like
> 'pp3', then you probably won't have any intuition about it.
>
> Writing good software is about conveying intentions, and part of the way we
> can do that is relying on what other people already know. Which means using
> familiar names and operators when it is helpful to do so.


Keep in mind that such intuitions often have a formalization.  We have an
intuition about what "map" means.  When we dig deep and try to write down
what that intuition is, the following appears:

   map id = id
   map (f . g) = map f . map g

Now that we have an operation and laws, it is reasonable to use a typeclass.

To me, typeclasses are at their best when you have a real abstraction to
encode.  If you are having trouble using a typeclass and need C++-style
ad-hoc overloading, it's likely you are trying to encode a "fake"
abstraction -- one that has only linguistic, rather than mathematical
meaning.

Haskell is not an isolated linguist.  Her low tolerance for vagueness
strikes again.

Luke


>
>
>  I would
>> consider any book which is hard
>> to read because of that badly written. Things are quite similar with the
>> code.
>>
>
> I consider the current state of affairs quite poor: namely, abuse of type
> classes and alternate names and operators that aren't very suggestive, but
> were chosen purely to avoid conflicts.
>
>  Programming language should be easy to reason about for both computers
>> and humans. Compiler should therefore disallow programming style that is
>> inaccessible for potential readers. Want to overload something? Well,
>> use typeclasses to be explicit about it.
>>
>
>
> Type classes were not designed for name overloading. They're designed to
> factor out common patterns in programming. You shouldn't use a type class
> just because you want to use a name or operator.
>
> And as I said before, if you want to "disallow programming style that is
> inaccessible for potential readers," then you should disallow the current
> state of affairs.
>
> Regards,
>
> John A. De Goes
> N-BRAIN, Inc.
> The Evolution of Collaboration
>
> http://www.n-brain.net    |    877-376-2724 x 101
>
>
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.haskell.org/pipermail/haskell-cafe/attachments/20090214/772d2b80/attachment.htm