autoconfiguring kernel ?

Sheldon E. Newhouse (sen@math.msu.edu)
Wed, 24 Apr 1996 09:03:24 -0400

Excuse me if this is not the proper place to post this. I had some ideas
for possible future development, kernel configuration, etc. Perhaps they
have already been discussed. If so, again please excuse me.

1. How difficult would it be to have a self-configuring kernel as an
option? What I imagine is that one runs a program to probe the existing
hardware, load the appropriate drivers, configure the appropriate
modules, etc., give the developer a report of the proposed configuration
with options to change things, and then make an appropriate kernel. It
may be, e.g. that certain configurations of drivers work better together
than others. At present, each system administrator has to try this or
that to tune his or her system. This should always be one of the
options, but some people might like to get the benefit of others
experience with auto-configuring kernels.

2. An extension of the above would be autoconfiguring installations and
upgrades. Again, only as an option. One of the problems with other
attempts to automate is that they take control away from the
systems-administrator. I don't think that this should not be done
without very good reasons.

I would be interested in follow-ups to these ideas.

Thanks,

-sen

---------------------------------------------------------------------------
| Sheldon E. Newhouse | mail: newhouse@math.msu.edu |
| Mathematics Department | www: http://www.mth.msu.edu/~sen |
| Michigan State University | telephone: 517-355-9684 |
| E. Lansing, MI 48824-1027 USA | FAX: 517-432-1562 |
---------------------------------------------------------------------------