[Thus spake Luke Mewburn ("LM: ") 11:52pm...]

LM: I currently don't see the major benefit in providing different
LM: end-user configuration files based on MK<var> variable settings.

Agreed.

I *do*, however, see the benefit in conditionally modifying the end-
user configuration file(s) based on MK<var> variable settings, by way
of a degree of preprocessing.

[Okay, so that's a semantic argument.  Sue me.]

I think the mtree stuff could stand to be either m4 or cpp input,
for example, from whence the final mtree def file is generated
prior to generating 'sets' and/or 'distribution'.

I'm quite surprised this has not been attempted/incorporated already.

				--*greywolf;
--
Support Open Operating Systems -- subvert the Microsoft paradigm.