I agree with much of what was said. I’m completely new to functional programming so it has been interesting to look at coding from this novel angle. I’ve found it a little hard at times to shake off my imperative coding habits.

So I have followed the examples and tried the exercises. One piece of code that just looked wrong to me was the find_sync routine on p85. This looks for three consecutive MPEG headers in the data. The issue I had with this is that it is basically the same piece of code embedded within itself three times. Not what I would call elegant! I was forced to rewrite this just for my own piece of mind. I added it to the example code and called it find_sync2. You can find it here:
http://oxtremists.co.uk/svn/erlang/mp3_sync.erl
any comments on this code would be readily accepted. I did test it on a real mp3 file and it seemed to do the same as the original.

Something else I started was a sudoku solver program. The main problem I hit with this was in deciding how I should store the data representing a grid. Initially I started out with a list, as that seems to be the standard collection type in use in Erlang. But lists are not really random-access collections. There is no BIF to allow you to replace the Nth item with M for example. I also started noticing that the list processing BIFs that there are seem to be rather inconsistent. So lists:sublist which gives you the first N items in a list takes the argument list (List, N) but lists:nthtail which does the reverse and gives you the items following the first N items takes the argument list (N, List).

So how about tuples? There’s a BIF to replace the Nth element with a different one, but there are very few other BIFs available to do anything else. Finally I found that there is an array module (not mentioned in the book) http://erlang.org/doc/man/array.html should I be using that? Basically, I’m confused. Obviously I could do it using any of these, but I don’t want to go re-inventing any wheels on the way….