Sebastian Sylvan wrote:
Perhaps a single largish application could be the "end product" of the book. Like a game or something. You'd start off with some examples early on, and then as quickly as possible start working on the low level utility functions for the game, moving on to more and more complex things as the book progresses. You'd inevitably have to deal with things like performance and other "real world" tasks. It might be difficult to find something which would work well, though.
This again reminds me of 'Write yourself a scheme in 48 hours'. It is exactly this approach, albeit on a far less ambitious level (tutorial, not book). You end up with a working scheme implementation; ok, a partial implementation missing most of the more advanced features, anyway, you get something which /really works/. I have spent quite some time adding stuff that was left out for simplicity (e.g. non-integral numbers), rewriting parts I found could be done better, added useful functionality (readline lib for input), improved the parser to be standard conform, added quasiquotation, etc. pp. Had lots of fun and learned a lot (and not only about Haskell). Cheers Ben