[developers] extension to the REPP sub-formalism

[Stephan Oepen]

hi again, mike,

> Do you have a strong case for exempting masked regions from the normal characterization behavior?

far from it; that would in fact seem counter-intuitive to me just now.

i have a sense i quite likely wrote that footnote you cite, but i also realize i do not currently have the full rationale behind that conclusion available (and do not feel like doubting it).

in any case, thanks for the refresher.  with this assumption, i now see why duplication or swapping around of masked regions is problematic.  i sometimes worry that REPP may be a bit like, say, awk(1): lurking edge cases that the original inventor(s) did not anticipate make a sound and complete implementation harder than first meets the eye :-).

all best, oe