cvs commit: src/sys/i386/include param.h

[Julian Elischer]

David O'Brien wrote:

>On Wed, Sep 01, 2004 at 01:38:20PM -0700, Julian Elischer wrote:
>  
>
>>I'm rapidly coming to that conclusion myself..
>>
>>breaks all over the place because so many places pretend to be the 
>>kernel when they are not.
>>
>>    
>>
>
>What it really means is we haven't provided a sufficient API for userland
>to get a kernel data that obviously is needed.
>

yes.. it's tricky..
should we define a "_KERNEL_INTERFACE" or something?
how about _REALLY_IS_KERNEL
which defines _KERNEL
so if you just need the interface you define _KERNEL
but include files cna protect REALLY internal stuff with "#ifdef 
_REALLY_IS_KERNEL


suggestions welcome.

>
>  
>