Wolfgang,

> Again, I’d still like to have a guarantee that failures will signal
> an error when the program is compiled/run in “debug mode”. Is there
> a situation in which this would impose a performance penalty on
> non-debug-mode programs?

Guaranteed debug mode errors should be fine—easy enough to toggle them
off. I agree that at least that (errors in debug mode) should be in the
SRFI. I encountered this debug silence myself just now when testing on
different implementations. No errors even in debug mode is strange.

I'll reword the spec with recommendations to error in debug mode,
leaving the non-debug part up to implementors. Does that sound alright?

Thanks,
--
Artyom Bologov
https://aartaka.me