Ketil Malde wrote:

Let's see. It all depends on what you put into the chapters, of course, but you seem to jump directly from simple concepts into Haskell for web servers. Wouldn't it be better to push this further out, until at least some of the syntax has been explored?

The second chapter is meant to be a brief overview of some of the ways you can execute Haskell code. For a programmer familiar with some of the more common languages (C++, Java, Perl is the kind I'm expecting), showing them right off the bat that Haskell *can* be used in web servers seemed like a good idea. It met two goals: a) this list's objective in spreading the good word about how useful Haskell is :-) b) the objective of "Nutshell" books to let experienced people transfer their existing skills as quickly as possible.

...

From the headlines, one might get the impression that type declarations are only important when the compiler can't figure it out itself - I'm sure the actual text will give a different view, no?

Of course - I have something of an informal writing tone when I'm writing for education, and some of the headlines reflect that (I mean, really, "hash"? :-) ) But part of what I was trying to do was to get some key concepts across to two different audiences of programmers: a) the Perl programmers, who have types but generally don't ever write them down. These people will like type inferencing - it will seem more natural to them. b) the C++ and Java programmers. They'll be just fine declaring types - showing them how to get strong typing without necessarily declaring the types explicitly will seem nifty cool to them (it did to me, ten years ago :-) ) I'm not about to get into any of the old arguments about static vs dynamic typing or that stuff - been there, done that. Haskell is a strongly typed language, and that will be apparent in the text.

The cookbook looks great, and it's certainly something I'd buy the

Do you think the cookbook would be better as its own book?

I can, of course, think of many kinds of example programs I'd like to see - how about a GUI oriented program, for instance?

Actually, last night after sending the announcement, I added exactly that :-) Years and years ago, Paul Haeberli of SGI implemented this cute little hack of having a brush attached to the cursor with a spring - real simple - moving the mouse changed the velocity vector of the brush, with settable strength and inertia and such. It made it insanely easy to produce a swooping, calligraphic stroke. Haeberli did it in C to OpenGL - using Fran, it would be perhaps a one-liner. Scott Snibbe, then of Brown University, did another cute hack that showed up at SIGGRAPH later under the name "MotionPhone". He set up a loop of screen buffers that ran at a constant frame rate. As you drew, your mouse left behind moving streaks. Faster motion, longer streaks. This would be another one-liner with Fran (Scott originally did it in C to Pixrect, Sun's ancient direct screen access API). I figured those kinds of examples showed off the power of Haskell.

Perhaps you could finish it off with a "real" application, say a GUIed JPEG viewer, where you sort of string it all (the JPEG codec, a C library interface for something or other, and GUI stuff, etc.) together into a whole?

This morning I've been thinking about data visualization from an SQL DB....

On the whole, I think it looks good. Best of luck!

Thanks! Brook