Let's skip past whether OpenDylan is ready for production use ("No") and pretend it is.
I'd absolutely use Dylan over those languages in at least some contexts.
Dylan hits a sweet spot between performance and flexibility. Even the existing OpenDylan ought to perform better than most of the languages you named, for example. But the level of flexibility is far greater: you get full-blown macros (Clojure, Common Lisp, Scheme), optional static typing (Racket), a really pleasant syntax that doesn't make me use a pile of parentheses, has a great object system based on multiple dispatch...
There's an awful lot to love there. The only reason I don't use it is because the compiler implementations have historically been very poor, and the library compatibility worse. If this project goes well, I'd definitely take another look.
Well, let's face it: There's only one compiler actively maintained. And its license is MIT!
NB: written in Dylan itself!
This even has an IDE (on Windows only), but has some problems with error reporting, producing too much output while compiling and a limited set of libraries.
The only way to improve that is to try it out, and contribute libraries!
I'm happy to look into compiler bugs - that is something I'm used to do over the last years ;)
Given the rather small community size (from what I know), wouldn't it be the best survival technique to port the compiler to JVM, to have at least some usable real world libraries at hand ? I can hardly imagine anyone starting to use a complex typed system like that without reliable and proven libraries from the real world. It just wouldn't be worth the effort of playing around with it.
I'd absolutely use Dylan over those languages in at least some contexts.
Dylan hits a sweet spot between performance and flexibility. Even the existing OpenDylan ought to perform better than most of the languages you named, for example. But the level of flexibility is far greater: you get full-blown macros (Clojure, Common Lisp, Scheme), optional static typing (Racket), a really pleasant syntax that doesn't make me use a pile of parentheses, has a great object system based on multiple dispatch...
There's an awful lot to love there. The only reason I don't use it is because the compiler implementations have historically been very poor, and the library compatibility worse. If this project goes well, I'd definitely take another look.