07:11 <leon-p> huh, just found out that if you use chicken scheme, you can compile the scheme program into an executable binary, without loosing the ability to load another scheme file at runtime which can re-define functions and globals

07:12 <leon-p> I wonder if that would be the better choice here over guile

07:13 <leon-p> the main advantage over guile is that on first startup guile byte-compiles the program, which takes a while

07:13 <leon-p> speed-wise they should be reasonably close, since guile is quite well optimized and even JIT's hot bytecode

07:34 <leon-p> but chickens FFi seems more cumbersome...

07:46 <leon-p> yeah I think I'll stick with guile for now. if I ever want to switch, all I'd need to do is re-write the bindings and replace one or two guile specific functions, otherwise scheme code seems to be fairly portable between both

09:05 <leon-p> ok that idea has haunted me the entire bus ride to the lab, so I think I'll also provide chicken binds and try to write the reference WM so that it can be used by both, dammit

09:09 <LarstiQ> leon-p: how long is the bus ride? :)

09:54 <leon-p> ~20 minutes, plus the ~5 minutes between two busses I spend at the bakery

10:01 <leon-p> unfortunately, there seems to be no good way to detect which scheme implementation you are running, so unless I am willing to use a preprocessor (ew) I'll probably have to shelf that idea

11:24 <leon-p> ok, srfi-0 technically allows to detect scheme implementations, but chicken seems way to limited to be used comfortably (no define*, no nested defines, etc.) so I'll wait until r7rs probably before considering that again