LSB1.1: /proc/cpuinfo

[Doug Beattie]

Alan Cox wrote:
> 
> > Then, in general the specification should state that any options not
> > documented, but found to be present in any command/function, should
> > not be used as they will not have been tested and results cannot be
> > guaranteed.
> 
> The LSB doesn't define what happens when you use a command or library
> function that is not in the LSB. If you use glibc internal functions you
> will get burned (especially if the folks trying to replace glibc with
> something much lighter actually succeed)

So are you saying that we should go through the process of adding
specifics to each command for the next version of the specification so
such problems can be avoided?

What would you recommend?