[Haskell-cafe] Why Haskell is beautiful to the novice

[Tobias Dammers]

On Sat, Aug 29, 2015 at 01:41:52AM -0700, Alexey Muranov wrote:
> IMO, what attracts a big part of kids to programming is the possibility to 
> program side effects.

What attracts them are the effects. Implementing them as side effects is
largely a choice the designers of their language have made for them.

> It seems to me that Haskell takes a big care to "seal off" the side
> effects and does it in a nontrivial way. This may complicate
> introduction to programming.

Haskell simply refuses to allow side effects at all (aside from
loopholes like unsafePerformIO). Making effects explicit complicates
some things and simplifies others.

> Telling the kids to "just use the IO monad and don't worry want a
> 'monad' is, it is just a magic word, it comes from Category Theory,
> don't try to understand, just follow me" might not be a good way to
> teach.

IO being a monad and IO being the type we use to describe real-world
effects are orthogonal concerns. What we should teach people is that:

- IO is the type we use to describe real-world actions
- IO actions can be combines to construct complex effectful programs
  from simple building blocks
- Many of the ways in which we can combine IO actions follow the Monad
  pattern, just like we can combine Strings in ways that follow the
  Monoid pattern.
- You don't need to fully grasp the Monad pattern in order to use it
  with IO; just get used to how monad operations work for IO and go from
  there. The more general intuitions will follow in due time.

> I've seen assembly language mentioned here, and what attracted me to it, 
> when i was a kid, was the possibility to program "side effects" explicitly. 
> Even if i could not observe those side effects, like change of a register 
> value, directly, they could be tested indirectly.

The value of teaching assembly language is that it's close to the metal,
and thus suitable for a learning trajectory that roughly follows the way
programming language design has unraveled over the past 60 years or so.
Assembly language is, in fact, an abstraction already: by mapping
mnemoics to machine instructions, we make them more palatable for
humans. The observation that instead of looking up the bit patterns for
machine instructions in a printed manual ourselves, we could have the
computer do it for us, is what I consider the spark that bootstrapped
the entire field of programming language design.

That doesn't mean, however, that this is the only possible trajectory -
it's a thorough one, and a very suitable one for people like me who want
to understand it all before moving to the next level of abstraction, but
most people, I realize, aren't like me.

> 
> Alexey.
> 
> On Saturday, August 29, 2015 at 8:10:58 AM UTC+2, Donn Cave wrote:
> >
> > quoth M Farkas-Dyck <stra... at gmail.com <javascript:>> 
> >
> > > ... I see much praise of 
> > > Python, while Haskell mostly performs better, is less verbose, and 
> > > catches type errors. Worse yet, I see counsel to learn it as a first 
> > > language. 
> >
> > Sure - "Programming for Everybody" is practically a Python trademark! 
> >
> > It is kind of embarrassing when Haskell enthusiasts see Python as a 
> > better language for beginners.  But in either case I think we'd expect 
> > only a fairly superficial treatment of the language, right?  I mean, 
> > for example, back in the day, one of my colleagues picked up Python 
> > for random minor utilitarian purposes, and when I talked to him he 
> > hadn't used classes for anything, so for him it was only incidentally 
> > OOP inasmuch as some of the built in functions were addressed as object 
> > member functions.  A beginning student doesn't need to learn OOP in 
> > any kind of depth.  He or she would need to learn about the IO monad, 
> > but maybe not monads in general.  I suppose that might somewhat limit 
> > one's potential appreciation of Haskell's beauty, if we're still 
> > talking about that. 
> >
> >         Donn 
> > _______________________________________________ 
> > Haskell-Cafe mailing list 
> > Haskel... at haskell.org <javascript:> 
> > http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe 
> >

> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe


-- 
Tobias Dammers - tdammers at gmail.com