[lpi-discuss] Principles for creation of exam objectives

[ross e. brunson]

On Fri, 2005-07-29 at 09:34 -0700, Mark Miller wrote:
> I welcome your input and thoughts. There are a number of innovative ways
> to test such variations available. Unfortunately, either the vendors do
> not support such innovation or the costs are more than the benefit is
> worth. 

See the other thread about the BSD task analysis, there are many good
polling options out there, is it worth putting the gnarly topics into a
quick and broad-based poll to get some new opinion DNA?

> This realization is at the core of our approach in fact. To the greatest
> extent possible we "boil down" the common features across major
> distributions and go with that. This can be exceedingly difficult at
> times and impossible with some. 

Yeah, Amen to that.  I had to try to do that with the SAIR exams, we had
9, count 'em 9 distros that were considered to be the base for
questions, and I probably got most of my grey hair during that period...

> For example there seems to be no logical
> way to find common ground on starting and stopping services between a
> Redhat based system and a Debian based system. 

HUH?  How about the full path to the darned things?  Last time I checked
there are still /etc/init.d and /etc/rc.d/init.d locations, and if
someone wants to have a vendor and distro neutral cert, they better know
at least that much!  

I even made a little script for my SUSE boxes, named it service and put
it in the /usr/local/bin, helped me a lot.

#!/bin/sh
/etc/init.d/$1 $2

> Both CAN support the
> Debian method but typically Redhat administrators will use a different
> method. It would be unfair to expect them to know the lower level method
> because they never use it! 

Sorry, if they don't want to know the lower level method, they shouldn't
have the vendor/distro-neutral cert.  Price to pay, to know how the
things really work, not just what cute little command and GUI to use.