On 2019/04/19 10:04, Simone Piccardi wrote:
Il 16/04/19 14:45, Mark Clarke ha scritto:
I would suggest that its not an either or approach. We could have a
part that is multiple choice and a practical part. The practical part
doesn't have to be under exam conditions. It could be a task like
write a bash script that does x or some other assignemtn. The student
is given 2 days to do the task and submit the script/assignement and
the testing can be automated.
And how do you avoid having the student getting "help" from a friend?
That's an excellent point.
Another is how will an automated tester account for every variation that
the candidate might have or do? Perhaps a candidate might validate an IP
Address (sensible) and naturally uses Python with netaddr. Automated
testing is likely to fail and the assignment, whilst correct, is marked
wrong. Now manual intervention is needed and that means salaries. The
cost of an exam just multiplies many times.
I've stayed out of this current discussion as it rears it's head every
few years and never goes anywhere. Such discussions are tiring.
Someone earlier mentioned the perception that hands-on testing is
better. I very much agree that it is a perception. It might not be true.
So what is hands-on testing good for? It's great for testing if a
candidate can perform a series of predetermined steps in response to a
given situation to produce a determined result. Hence why we test
student pilots with it. And electricians, scuba divers and almost every
action a sailor will do on the job (when sailors can't pass these tests,
other sailors die).
It's why RedHat, Cisco and SuSE use practical tests - those distros
provide specific tools to do specific functions and the candidate can
rely on the tools to be present and work correctly. To do task X on RHEL
regarding selinux, RHEL provides a tool, and it will be present on the
test machine. The candidate is required to show they can drive the tool
to produce the result RedHat demonstrated in the course.
In truth, this has very little to do with results, it has everything to
do with the tool and how it is used, and the result is a side-effect.
RedHat never puts anything in their low and mid level exams that is not
covered in sufficient detail in their course materials, to do so would
be very unfair. You can't expect someone to perform a task they were not
taught how to do.
If we look at LPI's mission, we see that it is to a large degree exactly
opposite to the above. LPI is not about RHEL tools, it is about the
candidate proving they understand Linux systems within the scope of the
level tested. Because the scope is not bound to a specific distro or
release, testing has to be done on a somewhat abstract, conceptual
level. There is nothing wrong with measuring the extent of conceptual
knowledge and this is what LPI does.
Testing conceptual knowledge is not inherently better or worse than
practical testing, they are simply different. Both have their place and
they are answers to different questions about candidates and should not
be conflated.
--
Alan McKinnon
[email protected]
_______________________________________________
lpi-examdev mailing list
[email protected]
https://list.lpi.org/mailman/listinfo/lpi-examdev
