Hacker News Reader

Comment by andrewcooke | Hacker News Reader

[disclaimer haven't used haskell for a decade, so this is pure speculation and not intended as any kind of informed criticism.]

is it possible that now would be a good time (just because it is a time when things seem to be changing) to simplify / clean up haskell? what i am wondering / asking is whether more recent additions to the language (type system) are generalisations of earlier features. in which case, maybe the earlier features could be dropped?

is anything like that possible or reasonable? it's a fairly old language, which has evolved a lot, and it stuck me it might be a question worth asking... (something like python 3 i guess, but a bit more meaty)

View on HN

I believe what you are thinking of is called "Haskell Prime". Haskell' is an effort to evolve the language standard of Haskell, and my impression of it has always been that it's rather analogous to Python 3, except that Haskell seems to evolve much more quickly than Python did, and in ways other than the accretion of new language builtins and keywords.

If you've not used Haskell in a decade, I think you'll find the language has changed significantly since then, but that sometimes the language disguises itself by choosing conservative language defaults that are maximally compatible. If you want to see how far Haskell has come it will require you to use a number of LANGUAGE pragmas to get the benefits of the most recent innovations.

Haskell Prime is not like Python 3 in the sense that there are very few (and no in practice) breaking changes to the language. Haskell Prime standardizes stuff that's already there.

> sometimes the language disguises itself by choosing conservative language defaults that are maximally compatible

Also, they don't like to change their name very often.

Seriously. It matters. Lisp has been Lisp for the past six decades or so. Therefore, someone who learned Lisp in the 1970s thinks that Lisp Is Lisp and therefore hasn't changed, except maybe for a few dialects. (After all, C and Java are just Algol dialects, right?)

The fact Clojure actually has a new name is likely more important to its success than anything else the language creator did for it.

"I don't know what the technical characteristics of the standard language for scientific and engineering computation in the year 2000 will be... but I know it will be called Fortran." John Backus

In 2000 Backus was still right, but not so sure about 2020. Much more of science has moved to other languages, indeed science that uses computation has got broader too, using R, perl, python, all sorts of stuff.

Now of course all communication links are still called ethernet, despite the lack of connection between 10Mb and gigabit, almost nothing in common bar the name.

> is it possible that now would be a good time to simplify / clean up haskell?

No, that is usually the worst time. Don't pile change on change. That doesn't make it easier.

Reply on news.ycombinator.com