@Jon, I would like to see more type inference added to Java, something along the lines that Scala or Fortress does. I am not sure about the extensive inference common in functional languages; I find the error messages poor. In fact I have suggested extensions including more inference already:
http://www.artima.com/weblogs/viewpost.jsp?thread=182412 Changing subject: an example of the expression problem that is pertinant to discussing computer languages is if you write a parser for a language and then want to expand the language at a latter date. In this case you might need to expand the number of expressions to cope with new language features. I note that there is currently much work on Haskell to remove this 'closed world' limitation from that language, so I assume I am not alone if finding 'closed worlds' a problem. Cheers, Howard. On Nov 12, 11:07 am, Jon Harrop <[EMAIL PROTECTED]> wrote: > On Sunday 11 November 2007 20:22, hlovatt wrote: > > > Most of your arguments seem to be around short syntax not about the > > semantics of the language. If shorter syntax is of such importance > > then add some sugar, > > The lack of features like inference and automatic generalization is the source > of verbosity here and you can't address that with sugar: it is a deficiency > in the language. > > > > > Moving back to examples, but this time semantic in nature rather than > > syntactic. Suppose that I was writing a program and using the symbolic > > math library Exp that was from a supplier, I didn't write it. Now in > > my program I want to add Pow, but even if I have the code for Exp it > > isn't a good idea to change Exp since I will have introduced a fork > > (you don't even want to recompile Exp - just use a binary). With > > multiple dispatch you can: > > > public class Pow extends AbstractExp { > > public final Var var; > > public final int pow; > > public Pow( Var var, int pow ) { > > this.var = var; > > this.pow = pow; > > } > > public String toString() { return var + "^" + pow; } > > public final static Exp d( Pow p, Var x ) { > > if ( p.var != x ) { return new Lit( 0 ); } > > return new Mul( new Lit( p.pow ), new Pow( p.var, p.pow - 1) ); > > } > > } > > > This is hard (I wouldn't say impossible since I don't know ML well > > enough) in the closed world view of ML semantics. > > > In summary: introduce some sugar for common cases; but don't go > > overboard, because the semantics are the really important thing. > > Jacques Garrigues paper on code reuse describes one way of doing that in OCaml > (using polymorphic variants) but I've never encountered that problem (the > expression problem) in practice and all solutions to it are necessarily > invasive. > > -- > Dr Jon D Harrop, Flying Frog Consultancy > Ltd.http://www.ffconsultancy.com/products/?e --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "JVM Languages" group. To post to this group, send email to jvm-languages@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/jvm-languages?hl=en -~----------~----~----~----~------~----~------~--~---
