Some say it's the equivalent of C #, or the equivalent of Scala's improvements to Java, and I can tell you that Swift is not the language. It's more like a toy language that makes a quick prototype.
The call for Apple to release the next generation of Objective-c languages is long overdue, with tens of thousands of-word OS X reviews Ars Technica wrote two years ago that Apple should do a language, objective-c without C and use more modern features and more Short API, and support for automatic type deduction, see OS X 10.8 Mountain lion:the Ars Technica Review.
But Apple apparently didn't take the language seriously. It's obvious that they don't want to replace the OBJC at all. And in Apple's style, if they treat Swift as OBJC's future, there will surely be one or two IOS/MAC applications that use swift rewriting. (for example, when Apple hacked Carbon, the first thing to do was rewrite the Finder).
From a language perspective, Swift has too many drawbacks:
-its standard library is too poor, only Dictionary, Array and String and several data structures, not even a Set. Want to use something else? You still have to use the OBJC.
-it does not have any support for asynchronous/synchronous programming. It cannot be bundled with GCD, so any serious Web application cannot be written at all.
-it does not have any protection mechanism (private, protected), all the method and variable are globally visible, which obviously can not meet the needs of serious projects.
-only interact with OBJC. If you want to use C or C + + seriously, you have to write OBJC wrapper.
-There is no way to catch Objective-c's exception, if once out of the exception, your App will immediately collapse.
Chris was able to clearly see his short-sighted in designing the language. This is normal, he is a compiler, not to do language research. So the language that this door claims to support functional programming is not functional at all:
-Only supported in lazy, closed, and simple function language functions (such as map, filter). You can do it in OBJC.
-No pattern matching
-Type deduction is almost the same as c++11, without the use of Hindley Milner.
From these features, Apple's positioning is only for the OBJC headache developers to reduce the learning curve of the toy language introduced. Each of the previous five items is basically fatal, so Apple's own projects are not used. It is possible to do an early proof of concept prototype.
I can imagine that the implementation of Swift is very simple (LLVM is very mature, writing a new language is more and more simple, such as a few steps to be able to make a small language: 1. Kaleidoscope:tutorial Introduction and the Lexer, of course, it's possible that Apple will solve these problems in the future to make it really practical. But for now, Apple doesn't want Swift to replace Objective C.