On 2008-11-10, Roy Lanek <[EMAIL PROTECTED]> wrote:
>     instead of their heads.) In the hands of a great programmer (like James
>     Clark), it's astounding what can be done in a few lines of Haskel, and
>     it's interesting to compare the amount of Java code it takes the same
>     programmer to implement the same algorithm.

Yep, you can do incredible things in 10 lines. But to think how to write
those few lines takes ten times as long as the writing the corresponding
1000 lines of C. That approach doesn't scale to big projects, where you have
some other objective than an academic exercise in the language in question.
And Haskell just sucks for the moderately object-oriented approach that is
"obvious" for certain things. You have to build workarounds upon
workarounds, because the obvious approach can't be sanely done.[*] And it gets
very tiresome, if you have to try to come up with a non-obvious model for
everything in a big project. The massive threading approach that I proposed
might be workable and "functional" (as well as actually "message passing"
OO), perhaps even scalable, but it also would need better language support
to be convenient to use.

I am actually very very slowly working on something in Haskell, dunno if
I'll ever finish, and in all the time I've spent fighting with the language,
I'd already have finished the project in C. Actually many things that would
be a single line in C are tens of lines of Haskell... But the entire project
is really an academic exercise.


[*] That said, a lot of time should be spent on designing the 
API of important core libraries etc. that everyone gets to suffer
from, to avoid the crummy "obvious" approach, and maybe come up 
with something nicer.

-- 
Be an early adopter! Beat the herd! Choose Windows today!

Reply via email to