On-Line Survey 2006

LiC Comments

Many thanks to all students for taking the time to answer the course survey.

The results mostly speak for themselves. The course seems to be in good shape, overall satsifaction was one 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).

Specific observations:

• The change of hardware platform seems to have worked well. There was a high degree of satisfaction with the Slugs (compared to the u4600's in previous years). It also looks like there were no major glitches with using a completely new lab setup for the first time. Many thanks to Dave Snowdon and Dave Johnson, who did the hardware mods, and Godfrey, who did a lot of the software development.
• Someone complained bitterly about the libnfs documentation, and the distribution mode for patches. No-one else said anything, but this is definitely something we'll look into for the next time round.
• Someone found some of the milestones tedious (presumably the file system). This seems to be an isolated sentiment, but we'll think about it. The last thing COMP9242 aspires to be is tedious!
• The three-hour block and lecture theatre are a recurrent complaint, but hard to fix. We might try to pack less material in and rely more on students reading up on the recommended readings. Experience, however, shows that the majority of students won't do this.
  Each year we get comments about the time of the lectures. Students are pretty much split over whether they prefer them in the evening or afternoon. They were originally in the evening but moved forward based on student feedback.
  At least with the theatre, we could hold the course in a more relaxed environment at NICTA. Will consider this (but it means an extra 5-minute walk each way for the students).
• I'll think about the suggested “decision making in SOS” component. Talking a bit about the tradeoffs might be good. But keep in mind, one of the main strengths of the course (much appreciated by students) is that you get to find out for yourself, and thus really appreciate why some decisions are better than others.
• Re “harshness” of the prerequisite: I am always open to someone making a case, especially based on excellent programming experience. But experience has shown that people who have only achieved a CR in OS, and have nothing else in their favour, have a high probability of failing, and generally don't get much out of the course.
• I had to smile about the one answer about what to scale back: “Virtual machines. They are an hot topic but are they worth a three-hour lecture?” These days there are whole conferences about VMs, and most of the actual OS work (as opposed to middelware crap) that is published in the main OS conferences these days is on virtualisation ;-)
• Lecture overlap with COMP3231, especially in security: Will try to focus more on special issues in the future.
• What one student describes as a “disconnect” between lecture (on caching) and project is intented. The lecture covered the general issue of caching and its relevance to OS, your hardware exhibited one particular aspect of this (but all you had to know about caching was indeed covered in the lecture).
• I'll think about the “reading group” suggestion.
• Complaints about spec being “too informal” are balanced by “don't over-prescribe”. I think they are basically right.
• The old discussion of are balanced by “should be 12UoC” vs “should be harder”. I still maintain that the complete project can be done in 2 weeks F/T (and this has been demonstrated by students several times). The secret is to be very disciplined and do no more than what the spec asks. The problem (although most see it as a feature ;-) is that the project entices students to do more. That's fine, but don's say I didn't warn you!

All up, that was a lot of useful feedback. Thanks to all those who submitted!

Gernot