linux-ipsec: to be or not to be, that is the config question

Doing some internal fiddling with the way the kernel code gets built, I ran into a problem we've encountered occasionally in the past -- the code currently won't compile with the debugging option off. That should be an easy fix for Richard, but it got me wondering. The following are some tentative thoughts, subject to reconsideration, offered for comment.

Given that we now have klipsdebug to turn debugging on and off, is there really a good reason to have a compile-time option for taking it out?

Unless there is a significant overhead for having it in but turned off -- I haven't gone and dug through the code for details -- I say this should be dispensed with. Having debugging facilities available is a Good Thing even in production code (and this isn't production code yet!), provided that the overhead they impose is small. We should not be taking them out without good reason.

Also, the fewer compile-time options we have, the fewer different ways we have to compile the thing to test them all. (Which is, of course, the reason why this problem crept in again -- for some reason we seldom have cause to compile with debugging out...)

This line of reasoning can be carried further. Is there really a good reason for having configuration options for each and every transform? An option which should always be enabled is just useless clutter. I suppose that a specialized application might want to custom-build a minimal binary with only the bits and pieces expected to be wanted (or expected to conform to Ruritania's cryptography laws). How likely is this to be a requirement?

We should think seriously about eliminating any configuration option that nobody in his right mind is going to want to exercise. Providing more options than necessary just makes everyone's life harder.

                                                          Henry Spencer
                                                       henry@spsystems.net
                                                     (henry@zoo.toronto.edu)