So with the announcement that Apple's Swift will be open sourced, I spent a little time this morning going through the language guide to get a sense of the language. Open source means multi-platform, so prior to that announcement Swift wasn't really of any interest to me. There's no particular ordering to the following thoughts.
>From a "modern language design" perspective, I see essentially nothing in Swift that is interesting. You can see the start of some features that originated in functional languages creeping in (pattern proto-matching, opt type, type inference), but not really all that many of them. The language has generics, but not polymorphism. Some variants of enumerations look like discriminated unions, and it appears there are no non-discriminated (indiscriminate?) unions. Swift continues Apple's commitment to ARC, though the ARC/GC debate remains (so far as I can tell) a holy war. Some renaming of familiar concepts (Protocols instead of Interfaces). So far as I can tell, Swift permits method and operator overloading, but not function overloading. This seems like a very odd choice. There doesn't appear to be a specification of how inference works. The language reference, for example, does not give type rules. There is pseudo-support for currying, but it seems oddly done. There's also a very odd handling of closures when they appear in the tail argument position. It seems strange that this wasn't generalized a bit. There is a limited constraint mechanism on generic types. All in all, Swift seems like a language designed by people who don't actually know the current state of the art. It seems very ad hoc, with partially implemented ideas all over the map. Buried in all of this, I'm sure there is a new idea or two that I should be aware of. Can somebody try to enlighten me? shap
_______________________________________________ bitc-dev mailing list bitc-dev@coyotos.org http://www.coyotos.org/mailman/listinfo/bitc-dev
