I would argue that mixing in inferred typing would be a little weird as
well. The only ML language I have worked with is OCaml (and very rusty
at that), but from my recollection, the inferred typing requires that
mathematical operators identify the type. So '+' for integer addition
and '+.' for floating point addition. Personally, I would rather just
declare the type and be done with it. Plus, as you said, inferred typing
makes type errors potentially hard to track down. Though to be fair and
to show my bias, I prefer dynamic typing. While I understand the
benefits of static typing, I much prefer the flexibility of dynamic
typing mixed with unit testing.
Some ideas to toss out:
- Named parameters
- Default parameter values
- Integrated Regular expressions
- I noticed you planned a sort of continuation support, but it would be
cool to have full closures (if the Flash VM can even do them...I admit,
I don't really know too much when it comes to this area)
- static initializers
- some sort of support for collection querying/sorting/subselecting.
This is really more of a library thing. Someone did mention Linq, and
while it is kind of neat for its common sql-like approach to querying
any chunk of data, a good library can be just as effective. (and really
needs to be there to translate the sql into anyway).
- I think some sort of language exensibilty (Lisp Macros, Ruby's
'meta-programming') would be neat. Really, just having some ability to
manipulate the program.
- Function Hooks (ala Ruby 2):
function foo() : Void {
trace( "in foo" );
}
function foo:pre() : Void {
trace( "before foo" );
}
function foo:post() : Void {
trace( "after foo" );
}
The hooks wouldn't really need return types since they should ever
return anything, but something like that. Also, hooks shouldn't be
callable (so no, foo:pre() )
_______________________________________________
osflash mailing list
[email protected]
http://osflash.org/mailman/listinfo/osflash_osflash.org
