> Being based on Clojure for good and bad also means being dependant on the JVM. {snip} The JVM situation (either open, supported and slow, or not supported at all) makes Clojure a no-go on this entire platform.
Maybe the solution is that we need a Clojure implemented on Guile.
Edit: That would make a number of groups of people happy:
* the free-software advocates who like Clojure don't like its current license (EPL)
* the folks who like Clojure but don't want to use the JVM or anything else Java-related
* the people who want to use Clojure on JVM but cannot tolerate its slow start-up time and/or high memory usage
* the group who would like to re-implement Emacs, but not in Elisp, Scheme, or CL
* (Edit 2) the Guile dev team, who would get an influx of new users
* (Edit 3) GNU folks, who would see more people exposed to GNU and Guile
Aside, not sure what the multi-threading story of a Guile-Clojure would look like.
I'm not arguing that Emacs should target the JVM but don't see how building it on yet another bytecode VM is going to fix any performance problems.
OpenJDK has terrible performance. Oracle JVM has terrible licensing and platform support.
For the "emerging" platforms such as ARM that means that anything based on JVM is inaccessible, unless you are willing to accept a 30 second+ startup-time for any JVM based application.
If you have any doubts about this, try getting Clojure and the appropriate tooling working on a Raspberry Pi. You will be tearing your hair out before having written a single line of code. While the rest of the Linux userland (based on C) just flows by at a nice regular speed.
The JVM situation (either open, supported and slow, or not supported at all) makes Clojure a no-go on this entire platform.
While I'm not saying we should limit Emacs to what runs well on a RPi, I think it's worth considering ARM as a critical platform to support in the future, and the JVM has no room there. At all.