On 12/13/06, Donald Bruce Stewart <[EMAIL PROTECTED]> wrote:
* Give tips on how to answer questions
+ Ok. we can put up an article here. Some suggestions:
- Solutions with unsafePerformIO should be discouraged (moreso
;)
I'd like to at least suggest a slight qualification here, based on
some personal experience. A colleague of mine learned Haskell after
having spent a while doing systems programming, and (with the danger
of completely misinterpreting him here), the (perceived) lack of
unsafePerformIO et. al. led him to conclude that Haskell was a good
high-level language, but a certain number of projects would at least
require you to write a fair amount of your code in C. Once he saw
some code that (safely) used unsafePerformIO - particularly in fps, as
I recall - he changed his mind. To this day, he hasn't written any
code using unsafe*, but is generally somewhat more interested in
Haskell.
So, I'd suggest that demonstrating both the use of unsafePerformIO
when the question is obviously trending that way and mentioning why it
is a bad idea -- perhaps with an example of something that will
obviously segfault or otherwise break type safety - is sometimes a
better idea than simply not admitting its existence.
Also, I strongly support the necromancy/IO analogies.
/g
--
It is myself I have never met, whose face is pasted on the underside of my mind.
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
