On Thu, 2006-08-31 at 10:43 +0100, Martin Percossi wrote: > I would just like to add one more comment: I believe John has commented > on the problems of haskell. I would just like to throw in my two cents. > John thinks that the biggest problem is portability, and by this I think > he refers to building the various libraries on different platforms.
Actually I mean building Haskell itself (of course than includes the core libraries). I'm sure building gcc is harder. However Ocaml builds fairly easily on many platforms. If I could be sure I could just 'get hold' of Haskell on a wide variety of platforms without much fuss, I'd be more tempted to rely on it. Once the compiler and libs are built, building pure Haskell sources shouldn't be an issue (and building bindings to C libs is of course always a pain). For example I complained once that my Ubuntu box ran a Haskell program (Ackermann's function actually) very very much slower than Ocaml, Felix, C or whatever .. I means Haskell is a purely functional language and Ackermann is a pure function: Haskell should have won the race easily. It turned out the Ubuntu (and thus Debian) package for the AMD64 wasn't 'registered'. [laziness] > I'm not saying it's insurmountable: if you sat down and thought about > it, I'm sure you could work out ways to provide an easy syntax to get > around the problem. I'm not so sure: a good simple syntax is hard to find. > I come from a C++ background, and for while now I've been looking for > the "perfect" language. My wish list is: > - higher order functions, currying, closures, all that good stuff > - easy tuples, lists, and arrays (I would even like sets!) > - STRICT language - if I really need lazyness, I'll have to do a bit of > extra work. I would be willing to consider "hybrid" systems though: > haven't had any experience with them yet > - a good type system: sorry guys, here's where I think felix falls down > - the type system strikes me as quite hacky. In what way? I know it is hard to describe, and I'm not disagreeing! The type system is based on Ocaml/ML, with various enrichments and restrictions. It has a set of 'half complete' features like constraints: these are experimental. > - more type system: inference, please! at the very least, type deduction Felix does bottom up deduction, but not inference, because that's quite hard when coupled with overloading. > - monadic code: it would be cool to have a subsystem which is > essentially monadic, but that feels exactly like odinary imperative code > (a la ocaml). So as a programmer, you could write referentially > transparent code alongside imperative code. Agreed. Felix does it the other way: functional code is embedded in procedural code. It would indeed be nice to be able to handle the dual case, and with a fairly seamless syntax. > Felix seems to do a great job on this, > but I think it went slightly overboard in attempting source > compatibility - you sacrifice some elegance in the type system and as > well it seems to complicate somewhat garbage collection. There are tradeoffs. Some lead to a mess, eg lvalue[] crap. > when I look at ocaml, I think I'm looking at road kill LOL! > Sorry if this mail is rambling, and I apologize in advance for any > offending remarks: I don't mean to pick a fight here, I'm just outlining > what I would like out of a programming language. Well, there's an opportunuity to do something about it: Felix is a community project. -- John Skaller <skaller at users dot sf dot net> Felix, successor to C++: http://felix.sf.net ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Felix-language mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/felix-language
