
On Mon, Jan 3, 2011 at 6:19 PM, Ertugrul Soeylemez
Alex Kropivny
wrote: Could something like code abstraction be done instead?
Haskell lends itself to solving problems in really generic, high level ways that reveal a LOT about the underlying problem structure. Through some combination of descriptive data types, generic type classes, and generic helper functions... You get an extremely clear problem description.
Example: https://github.com/amtal/snippets/blob/master/Key.hs (Haskell) versus http://siyobik.info/index.php?module=pastebin&id=543 (C++)
Clarity is a lot harder to score for, so you'd probably need to score things via votes. (Unless there's a way to measure how "generic"/high-level code is?) Such a site would fill a very nice role, that the programming language shootout definitely does not fill.
Currently the only way to figure out what "good" Haskell code looks like is to browse lots of blogs, and dig through hackage until you find beautifully written packages.
I really like this idea. New concepts in Haskell come up from time to time. Now if there was a competition for code quality and good ideas, they may become more frequent.
This could also go for problems, where you have to use some specific feature or extension, like scoped type variables, type families/functional dependencies, or even just typeclasses... (Remember, there was a deep list concatenation problem thread some days ago.) -- Markus Läll