It makes coding not easier to understand I think only faster... But of course that's very important too ;-)

Does anyone use concepts like that in its own code?

(BTW a bit of topic, but I'm looking for some java code that is reading a mailbox and puts all the received e-mail (including attachment) in an sql database... if anyone....)

My thought was about the same as Art-One. It's a great idea to speed up code, but it adds a level of complexity that seems to out-weigh the speed improvement.

I suppose if an API was put together that made the code that used the concept easier to follow, then it would be OK.

Interesting...

Pretty powerful stuff. For example, you could map over the entire web, and then reduce over the small set of docs that matched some criteria.

Deeper magic, as C.S. Lewis would say.

I think that it only looks harder to understand because everybody is used to the for loop... but after using the map for a while, you start seeing it even easier than the traditional loop, because you don't need to keep control of an iterator variable (the i for example), the array length, etc..
The good news is that with the new languages/frameworks it is getting more widespread... Ruby and Prototype uses this extensively..

Anyway, I think that for a compiled language like C, where you are taking care of every byte of memory used, it is enough to have function pointers.. they serve the map function very well (creating functions on the fly seems a bit overkill).. (disclaimer: big C fan)

As always, Joel's articles are great. He illustrates his point very well.. I just didn't quite get how useful the reduce function can be in such scenario..