Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon. Entire thread

I like Haskell

Name: Anonymous 2016-03-11 22:42

Haskell is a great language and I hope it becomes more widely used and the community grows. I want Haskell to keep pushing forward with FTP and Dependent Haskell and not draw back because "noobs" might find things hard to understand.

Name: Anonymous 2016-03-11 23:11

It's not the noobs that hate FTP, it's Haskellers that hate it. Attempting to belittle your opponents is a childish response to disagreement.

Name: Anonymous 2016-03-11 23:35

>>2
Specializing the Prelude to lists was a mistake in the first place. FTP is just fixing that mistake.

Name: IHBT 2016-03-12 9:47

I want Kmeth to leave and stay leave
>>3
Alternatively you could have used different names for folds in Foldable and let Data.List keeps the familiar names.
Why the fuck did they think generalization was a good idea in this case again? Everyone familiar with functional programming expects foldr to be the primitive recursion operator on list, no screw that, foldr should work specifically on list. It can be a specific instance of the generalized fold for all ADTs for all I care, but this foldr. This fucking foldr. Must. Work. On and only on. Fucking. List.

Name: Anonymous 2016-03-12 13:56

Ideally typeclass instances would be fully free, enabled by whole-program optimisation.
Then splatter everything with more general operations (e.g. length :: Sized a => a -> Int) to reduce the number of import qualifieds you have to do for regular modules like Data.Set.
And, probably most importantly, give haddock a mode that lists all instance functions for a given datatype, rather than just listing all instances and making you open 60 tabs to figure out if you can foldM_ or concatMapM3_ or whatever.

I guess FTP was a good start, but a bit of a half-assed attempt at incorporating all useful generalisations that didn't make it into Prelude.

Name: Anonymous 2016-03-12 15:37

>>4
A dogmatic opinion with no grounding in evidence. Please wait while I laugh in your general direction.

Name: Anonymous 2016-03-12 16:02

>>4
Everyone familiar with functional programming expects foldr to be the primitive recursion operator on list
No they don't. You've just made that up out of your ass.

Name: Anonymous 2016-03-12 16:22

>>4
Alternatively you could have used different names for folds in Foldable and let Data.List keeps the familiar names.

This is why Haskell is a fundamentally broken language, why can't two different types have the same method? In a good language like Java, a List and a Map can both have a isEmpty() method without belonging to some super type that presciently covered every use case from the dawn of its creation.

Name: Anonymous 2016-03-12 16:54

In a good language like Java
IHBT

Name: Anonymous 2016-03-12 19:29

>>5
https://prime.haskell.org/wiki/NumericClasses
Something like this would be nice too.

Name: Anonymous 2016-03-12 19:35

>>8
IHNBT

Name: Anonymous 2016-03-12 19:35

>>8
You haven't heard of typeclasses, have you?

Name: Anonymous 2016-03-13 7:42

>>12
typeclasses are like javaclasses, but without dynamic dispatch

Name: Anonymous 2016-03-13 9:53

>>13
And also with static dispatch which shines when combined with powerful type inference. Anyway, typeclasses easily allow two or more different types to have the same method.

Name: Anonymous 2016-03-13 15:50

>>4
Agreed. Kmett is a disgrace to the language.

Name: Anonymous 2016-03-13 17:46

>>15
What worse is that he even managed to convince spj lense is a good idea.
HASKAL's future as a language successfully survives while avoiding success at all cost was bright, and then this shit happened. Well at least it is now avoiding success even more than before.

Name: Anonymous 2016-03-15 1:15

>>14
Typeclasses can't share the same method name, so every class needs to be designed perfectly for its domain the first time, or you have to keep making up new names for the same thing so classes with overlap don't step on each other toes.

Any language in TOYL 2016 should be ashamed if it doesn't let you generalize existing interfaces in a backwards compatible way.

Name: Anonymous 2016-03-16 1:33

>>17
Next they'll be saying Haskell isn't ``real'' functional programming and FP actually means something like Coq. The when people have trouble with Coq, it's not ``real'' functional programming either.

Name: Anonymous 2016-03-16 8:06

>>18
Coq is a proof assistant.
The real functional programming language is Igdris

Name: Anonymous 2016-03-16 9:50

If your liking of Haskell lasts for more than 4 hours, call your doctor.

Name: Anonymous 2016-03-16 20:54

>>17
A ``method'' that could have two different types is badly-named.
Of course, good design is lots of small typeclasses that do a couple of things each (as with everything), which the standard library has mostly done right so far; this also makes it difficult to come up with a new type that can only sort of half implement a typeclass but badly needs one of the names.

Name: Anonymous 2016-03-17 1:10

>>19
Idris
https://github.com/idris-hackers/IdrisScript/commit/818cdf7e73f6b7f816e089bdae97a1a528bcad6a

Name: Anonymous 2016-03-17 6:58

>>22
#SoProgressiveAndTolerant #SmashThePatriarchy #BlackLivesMatter

Name: Anonymous 2016-03-17 7:10

>>22
#Dubbs

Name: Anonymous 2016-03-17 13:01

>>22
Dude said Igdris
Agda's thick, juicy big black gd penetrating deep into the fragile and delicate Idris

Name: Anonymous 2016-03-17 13:04

Progriders are sexually starved neets.

Name: Anonymous 2016-03-17 13:18

/prog/riders are sex

Name: Anonymous 2016-03-17 20:23

>>22
https://github.com/idris-hackers/IdrisScript/blob/master/LICENSE
Newer Posts
Don't change these.
Name: Email:
Entire Thread Thread List
All trademarks and copyrights on this site are owned by their respective parties. All comments and posts are the responsibility of their own posters.
- Tinychan + 2ch Mode + RSS -