On 08/11/2010, at 4:56 AM, Lenny222 wrote: >> Last time (whatever that means) we got a few interested people that >> ALL wanted something weird: one wanted classes, one wanted to >> do serious stuff on Windows which we could barely support, >> etc ..and I ended up doing 10 different things at once trying >> to accomodate them. > > I am not such much interested in purely theoretic concepts and ideas. > To me Felix seems good enough and only needs more libraries and > documentation to be more interesting for the real world.
Well, I am interested in "pure" ideas/theory because I think it helps get a nice clean "orthogonal" design ..as well as making the implementation simpler. For example there's "if it looks like a function and quacks like a function then it is a function". Don't try this in Ocaml: type t = | X of int let f g x = g x in match f X 1 with X i => print_endline i ;; It won't work, because X isn't a function, its a type constructor. Sure, you can do let X_wrapper i = X i in match f X_wrapper 1 with ... and that will work. Felix does this automatically: this works: union t = | X of int; fun f (g:int->t) (x:int) => g x; println$ match f X 1 with | X ?i => i endmatch; In fact, Ocaml is simpler and purer here, what Felix does is a hack (X really isn't a function). It's a very interesting question what to do with constructors of "no" arguments. In principle these do not exist. For example type boolean = True | False Really, these are constructors and have to take a unit argument (), but then you'd be writing (True ()) everywhere. So Ocaml, and Felix, hack it. And so in that case, in both languages, you have to use a wrapper to pass it as a function. This kind of issue IS important when you're doing higher order type of stuffs.. handling special cases is a serious hassle. Just consider C++ template with this code in it: return x; where x is type T and consider the substitution T->void. Originally you couldn't do this.. I think that has been changed because if you don't allow it you require two templates, one for non-void and one for the void case, just to handle the fact that return argument has to be used for non-void and plain return if the argument type is void. Maybe two isn't a bad number .. but if you have a template of 3 arguments you now need 2^3=8 templates .. just because of a non-orthogonal syntax. Same problem exists with const/non-const (and volatile) pointers, which is why the committee (stupidly IMHO) provided a way to access the "constness" of a type, to save the template-meta-programming hackers from massive duplication. Anyhow, this one is a non-trivial and important issue for Felix because Felix does not have any support for "const" or "volatile" .. despite the fact it is supposed to be designed to be good at binding to C. -- john skaller skal...@users.sourceforge.net ------------------------------------------------------------------------------ The Next 800 Companies to Lead America's Growth: New Video Whitepaper David G. Thomson, author of the best-selling book "Blueprint to a Billion" shares his insights and actions to help propel your business during the next growth cycle. Listen Now! http://p.sf.net/sfu/SAP-dev2dev _______________________________________________ Felix-language mailing list Felix-language@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/felix-language
