On Tuesday July 27, j...@yonan.net wrote: > I agree with you that tunnel device type and subnet vs. point-to-point mode > are mutually exclusive properties. > > My main concern is preventing an explosion in the size of the parameter > permutation space, where the code has to handle 4 cases of dev/net > permutations instead of only two. > > $ grep DEV_TYPE *.[ch] | wc > 59 328 2855 > > There are a lot of places where DEV_TYPE_x is referenced, and many of these > places will now need to handle 4 cases instead of only 2.
I don't think so. These places need to handle either DEV_TYPE or NET_TYPE. In no case is a 4-way test needed. Some bits of code are interesting in the dev type. Some in the net type. There is no overlap. See that patch.... NeilBrown
