School of Computer Science & Engineering
University of New South Wales
Advanced Operating Systems
On-Line Survey 2007
Many thanks to all students for taking the time to answer the course
The results mostly speak for themselves. The course seems to be in good
shape, overall satsifaction was at par with '06 and among of the
highest ever (but given the
small size of the class, the statisitical significance of this result
is low, so should not be overrated).
- The biggest gripe was the quality of the documentation. That's
one of the challenges of working with a research platform. In 2008
this will be resolved by moving to the commercially-supported and
well-documented OKL4 platform (this would have been too risky to do
- The second gripe that was repeatedly made was that the security
lectures were boring, and there was a significant sentiment that
they should simply be dumped.
That might have been an option this year, although I still think the
material is very important. In 2008, with only 12 teaching weeks,
security is being almost completely squeezed out of COMP3231 and
friends, so it really has to be taught here.
However, I take the comments to heart. I will revise the material
and try to make it more stimulating.
12UoC was a repeated comment... This won't happen, sorry. People
know what they are getting in for ;-)
A repeated comment about prerequisites was that there should be
flexibility. Of course there is! I am always open to give someone
permission to enrol if they can convince me that they are up to
it. But experience has shown that most people with less than a D in
OS struggle, and either just scrape through or fail. I want students
to have a good experience in this course.
More direct discussions of research papers: I'll try to do more of
this if I can create space for it (12-week sessions will be a
A few things were only brought up by one individual (not always the
same, of course), so don't seem to be major issues. I'll comment on
didn't like the exam, probably doesn't test what I've
Actually, my experience is that the exam marks are quite well
correlated with the project marks. Also, the project is assessed
on its own (and contributes 2/3 of the total mark!), so the exam
shouldn't just assess the same knowledge/skills. I believe that
the exam assesses the insights I'm trying to get across in the
Not a lot of feedback on the project:
Good point. The project assessment should provide
Start on time, more breaks:
The first one is up to the students, many come late (I was on
time!). But I'll be less tolerant of latecomers in the
I'll also do more breaks (which is easier to do if the lectures
start on time).
More info on assessment criteria
I actually think they are quite clear. However, I noticed that
the tutors were stricter than in earlier years and frequently
docked off marks. I adjusted a lot of the milestone marks, as
the emphasis for milestones should be on helping students
(sending them back for an improved version rather than marking
them down). Tutors will be better briefed in the future.
Let best team (of every milestone?) present their design:
Hmm, that would require putting much more effort into assessing
milestones. I'd rather spend time on helping students. However,
I strongly encourage discussion, and in previous years there was
a fair bit of this in the labs. You guys are old enough to
How are monolithic OSes built (eg Linux):
We did a comparative OS design lecture last year. Requires someone with
a lot of experience in design and implementation several
systems, not always handy. (In '06 we had this, and the person
is actually quite an engaging speaker, but the topic didn't
exactly get rave reviews.) Will see whether I can line something
up next time.
All up, that was a lot of useful feedback. Thanks to all those who
03 Aug 2008.