> Well, essential or not, we can argue about. hw.devconf is what I
> consider "closed territory" and therefore these functions shouldn't
> try to mess with it. There exists special programs which are good
> at handling that area. These functions really of any value only for
> the variables that sysctl(8) handles.
Good point, when I think about it i realize what you mean, why should
a user that run "sysctl -a" have to see every little variable that the
kernel exports with the sysctl call, in cases like this there certainly
would be better to have a dedicated tool to parse the data.
> The entire sysctl area is still being reconsidered and rearchitected
> quite a bit, and this particular interface should >>>NOT<<< be relied
> on for the future.
But if I whant to create an interface for iostat to retrive data from
the disk drivers to generate an extended disk statistic, should I wait
untill the interface is "reconsidered and rearchitected" or will
it be a minor thing to convert from the current interface to the uppcoming
interface?
/Mattias