"Next Generation" kernel configuration?

[M. Warner Losh]

In message: <4101DC69.9030309 at anobject.com>
            Jake Hamby <jhamby at anobject.com> writes:
: Conrad J. Sabatier wrote:
: > On 21-Jul-2004 Devon H. O'Dell wrote:
: > 
: > 
: >>I'm sure this will become another bikeshed, so I suggest whoever came
: >>up with the idea to put up or shut up. People are interested in
: >>solutions, not suggestions.
: > 
: > 
: > Agreed.  And the original proponent of the idea was me.  I just wanted
: > to see if there was any willingness to even consider something like
: > this before I went and did a lot of work for nothing.
: > 
: > Seems the general concensus is that most people are OK with the idea,
: > depending on the implementation.
: > 
: > I'll be quiet now until/unless I can actually come up with something. 
: > :-)
: 
: If you are looking to improve the current build process, here's an idea 
: someone could implement that would save a lot of people a lot of time...
: 
: My biggest annoyance with building the kernel, compared to Linux, is 
: that it insists on building all of the possible kernel modules, even 
: though I only want to build the ones that make sense for my hardware. 
: In Linux, despite the drawbacks of the menu-based config, it is nice 
: being able to easily specify Yes, Module, or No for most options.
: 
: The least intrusive approach would probably be to add a second config 
: file (e.g. "MYKERNEL.modules") which would contain only the names of the 
: modules to build in some make-friendly format.  You could then modify 
: config(8) to automatically copy this file, if it exists, to the object 
: directory where it would be included by the appropriate Makefile.  If no 
: .modules file exists, then it would continue the current behavior of 
: building all possible modules.

What's wrong with adding:

makeoptions  MODULES_OVERRIDE="a b c"

to your config file?  It is already supported.

Warner