Very often the type system assures safety to a point where some coding is done trough trial and error: -I think that this composition is right, let´s try it-... -oh, some missing type here, Let´s put it here and here-...-go on-... -The typechecker say t´s OK.  I run it and, well it works. Later  when I look at the code, I have to think twice to understand it. Sometimes I mix transactions, threading, IO and some other abstractions in the same line, things that I would fear much to work with in any other language. 

This playing thig is fine, because the typechecking cut a lot of dead branches in the entrophic tendency of humans to commit mistakes. For this matter I praise math as the hidden god that handles reality and haskell its prophet. but the resulting code is not a product of my intended design, but something made together  with the compiler, and sometimes it needs more than a look to understand it, even to myself.

2010/7/5 Ertugrul Soeylemez <es@ertes.de>

Felipe Lessa <felipe.lessa@gmail.com> wrote:

> On Sat, Jul 3, 2010 at 9:25 AM, Ertugrul Soeylemez <es@ertes.de> wrote:
> > Haskell provides a lot of low level glue like laziness, currying and
> > other very helpful language features.  But what is different in
> > Haskell is that it doesn't seem to provide any high level glue like
> > other languages do, especially when it comes to the IO world.  There
> > is a somewhat powerful module system, but nothing to bring modules
> > and the objects they define together in a consistent way.
>
> When I first read this paragraph, I thought: "STM to the rescue!".
> STM is one of the best concurrent world glues, IMHO.

I found that I get along with the basic concurrency constructs.  STM may
be handy in a few applications, but in none that I write.


> If you want, you may use Haskell just as you as PHP or C: just put
> everything in IO.  Your code will get uglier and the type system won't
> catch many bugs, but that's what we get when doing C or PHP, right?

Not really.  Even when programming in such a style, Haskell is much
safer than PHP with its braindead type system, and still somewhat safer
than C.


> > The problem with that approach is:  This makes my code harder to
> > understand for beginners.  Usually they can tell /what/ my code is
> > doing, because it's written in natural language as much as possible,
> > but they couldn't reproduce it.  And when they try to learn it, they
> > give up fast, because you need quite some background for that.  Also
> > sometimes when I write Haskell, my friend sits beside me and
> > watches.  When he asks (as a PHP programmer with some C background),
> > say, about my types, I can't give a brief explanation like I could
> > in other languages.
>
> I agree that it gets harder to reason about the code.  In fact,
> sometimes I stack monad transformers in the wrong order.  However, as
> Ivan says, if the feature is useful for you, don't be afraid of using
> it.  Beginners may have a hard time grasping the concepts for the
> first time, but that's only until they "get it".

I find monad transformers easy to reason about, and in most cases the
stacking order doesn't make a difference at all.  Just remember to
change the running function, too.  The problem with them is that
beginners learn them very late.


> > Yesterday I was writing a toy texture handler for OpenGL (for
> > loading and selecting textures).  He asked about my TextureT
> > type.  I couldn't explain it, because you couldn't even express such
> > a thing in PHP or C.
> >
> >  type TextureT = StateT Config
> >
> >  -- Note that this is MonadLib.
> >  -- BaseM m IO corresponds to MonadIO m.
> >  selectTexture :: BaseM m IO => B.ByteString -> TextureT m ()
>
> "It is the type of functions that may access and modify a state of
> type Config."

Then you need to explain "type of functions" and this explicitly
implicit "state" and why you have to do it that way in Haskell.


Greets,
Ertugrul


--
nightmare = unsafePerformIO (getWrongWife >>= sex)
http://ertes.de/


_______________________________________________