« He saw it too| Main | Gas Prices: A Perspective »
F# Blues
So far my experiment in programming in F# has been an unmitigated disaster. Not that I have killed myself trying -- I simply thought that you installed the language and everything would work. Turns out it is not-that-ready-for-prime-time.
So I've played around with it part time, while working on other things. I found out that although the web application wizard doesn't work, it can be made to work. The F# Web Library? That was a nightmare. First I had to build the code using the new version of the F# Compiler, then run some special linking code the guy had set up. One dynamic link library compiles fine from the command line and the second does with some extended twiddling. But wait -- even though it looks like the second one is compiled okay, when I link it in to the rest of the project the compiler barfs. The error message? Cryptic and unhelpful. Do the guys that write compiler messages take some kind of course in obfuscation?
I'm really excited about using continuations across the web-client void, and using ML to code both the server side and client side. But quite frankly if I were even working on a beta project I would have had to abandon F# by now. Somebody needs to do the work of integrating it better in with the VS IDE. Command-line tools are great, but it's a little much to expect the average coder to switch back and forth between heavy CLI mode, the web, and an IDE. Add to that trying to figure out the new language while you're trying to get files in the language compiling? It makes for a heavy load.
F# is sharp. I'm not giving up on it or ML (or OCaml). It just looks painful along the lines of learning C++ instead of painful along the lines of learning, say, Javascript. Seems like all the really cool stuff in technology is a pain to get to.
Or maybe I just like complaining. Your call.
1 Comment
Leave a comment
Hi,
if you're refering to the F# WebTools project (http://www.codeplex.com/fswebtools) than I'm the guy who started the project. I compltely agree that the project is currently more "research" project (showing that an interesting idea like that can be implemented in F#) rather than anything useful even as aplha version.
Thanks for the comments mentioning what makes the project so painful to use currently - I'll definitely try to address the issues you mentioned in future - I think that part of the problem is that the project currently relies on too many beta versions (some beta of VS Orcas, some version of F# compiler, though it isn't clearly said which..).
Also, the F# WebTools is probably not the best project to start with when looking at F# (though I believe that the idea behind it is very attractive), mainly because it is really very "alpha" version and also bcause the whole ASP.NET support in F# is still evolving. Finally, the project also uses the language (especially recently added features) in a slightly different way than intended by the authors (client/server blocks), so I believe that once F# becomes more stable language, we'll find some way to make the whole project feel like a more natural F# component.
BTW: I'm glad you have not killed yourself while trying it :-). It's good to know that there are people who are interested in the project and will have chance to look at it once it becomes more mature.
Anyway, thanks for the comments!
Tomas