Re: Haskell Foldable Wast

...

...

...

...

Henrik Nilsson writes:

Well, I would not characterise many of those who have raised concerns about this as "laymen".

With a couple of decades of Haskell and CS teaching experience, and a couple more of general programming experience, I don't consider myself a layman either (and I hope that does not come across as arrogant).

My apologies, Henrik, I did not mean to insult those who have considerable experience, and are certainly not 'layman'. Permit me to retract that remark.

Personally, if I wanted to introduce annotated numbers, I'd introduce a new type to clearly convey that idea. Yes, a bit more to write, but the end result is code that conveys the ideas behind it in a clear manner. After all, code is (broadly) written once, but read many times.

This is indeed the sanest approach, but sometimes I receive such tuples from libraries, and it can be convenient to treat them as annotated values.

As to making tuples functor instances, that can only be done by arbitrarily imbuing one of the fields with a special status. I have to ask: Why? After all, taking pairs as an example, the *essence* of a pair is that there are two projection functions, one for each field.

Maybe because we don't have type-level flip? Lens does solve this problem, by allowing us fold on any type variable we like (provided the appropriate folding definition is given), but type classes don't allow it.

As to tuples as instances of foldable: Why? There isn't any structure to fold!

Well, I've used the Traversable instance of ((,) e) countless times, and right now Traversable requires Foldable. I'd hate to lose both by removing one.

And allowing us to spend valuable teaching time on aspects of Haskell that actually might persuade our students to take Haskell seriously.

Isn't it possible to use a Prelude in your class that restricts Haskell to the parts you want to focus on? -- John Wiegley GPG fingerprint = 4710 CF98 AF9B 327B B80F http://newartisans.com 60E1 46C4 BD1A 7AC1 4BA2