[an error occurred while processing this directive]

<CENTER>
<H1>Project<BR>
Being the Fat Controller</H1>
<H2>Markers' Comments</H2>
</CENTER>


<SMALL>
<h3>Index</h3>

<UL>
<LI><A HREF="#location">Location</A></LI>
<LI><A HREF="#network">Network</A></LI>
<!-- <LI><AA HREF="#interface">Others to come ...</A></LI> -->
</UL>
</SMALL>

<HR>
<!-- INDEX END -->

<P>

<A NAME="location"></a>
<h2>Location</h2>

<P>Tutors' will have specific comments for hand-marking.

<p>The far majority of Location modules struggled with automarking.
A worrying number of submissions did not even compile. Many people
seemed to lack understanding of ADTs, failing to even export their
Location ADT, or failing to export other functions necessary for the
interface. Others went the other extreme and had no export restrictions
whatsoever. Markers spent considerable time attempting to fix modules
that would not compile. While the majority were fixed, sadly there were
a few left which simply could not be salvaged, and they gained 0 marks.

<p>Please read the specification! Make sure you export functions the spec
says you must export. Import ThomasLib! Lots of basics were missed out.
Usually we were generous and only deducted 2 marks (or even no marks)
out of 20 for having to fix a module which would not compile.

<p>Once working, there were many problems. Many modules suffered from
run-time errors, others got into infinite loops and were killed after
3 minutes (the time allowed for each test). Those that did not produce
errors mostly got low marks as well. Congratulations to those people
who got high marks. This was a very challenging assignment and it shows
real brilliance to have scored anywhere above 15/20. The highest order
of congratulations goes to Liang Ma, the only student to score 20/20.

<p>The main suggestion of the markers is that students should have focused
on getting the most basic cases to work. This would have earned a reasonable
mark. The most basic case (just returning a track where a light is on) would
have earned 8/20, which is more than most students were able to earn. The
following code, written by a marker, though extremely short and simple, would
have earned 8/20: <pre>module Location where

import List
import ThomasLib

data Location = Locate [Track] [Light] [TrackName]

maybeTracks ::  Location -> TrainName -> [TrackName]
maybeTracks (Locate _ _ trackNames) _ = trackNames

makeLocation :: TrackData -> [(Train, Track)] -> Market -> Location
makeLocation (tracks, lights, _) _ _ = Locate tracks lights []

updateLocation :: Location -> LightStatus -> ControlStatus -> Location
updateLocation (Locate tracks lights _) lightStatus _
  | all (== False) lightStatus = (Locate tracks lights [])
  | otherwise                  = (Locate tracks lights [newTrackName])
  where
    newTrackName = getTrackName (head (lights!!lightOnIndex))
    lightOnIndex = head (elemIndices True lightStatus)

getTrackName (Simple name _ _) = name
getTrackName (Fork name _ _ _) = name
getTrackName (Stop name _)     = name</pre>
(Note that all tutorials had some function equivalent to maybeTracks)

<p>In terms of remarks, you will need a specific reason to apply for a
remark. Before you take issue, make sure you have read through the
automarking log very carefully. If you know for a fact that your module
needed an extremely large heap for execution (we restricted the heap size to 512 MB, plenty
for any remotely normal module) then come and see us.

<p>In conclusion, the advice is to read the specification in detail, and to
start coding by making sure the module works fully and correctly for the most simple of
cases. Once they are covered, then expand to slightly harder cases, and so on.
If this course of action had been followed, 70% of students would have got around
10/20. As it was, very few broke that barrier.
<P>

<p>(alanl)</p>

<A NAME="network"></a>
<h2>Network</h2>

</p>

<ul>
<li>The average mark was 1.4 out of 5.</li>
<li>33 people managed to score 0 because their code <a href="http://cgi.cse.unsw.edu.au/~cs1711ass/faq-serve/cache/964.html">didn't compile</a>, or did compile but didn't work at all!</li>
<li>20 people managed to fail Test 1, yet still get more than 0 overall, by passing other tests. Test 1 was one train doing laps of a 4-piece circular Simple-track network. As there was only one train, no barriers and no forks, there could never be an accident - yet people still said there was! How you fail something like this and still pass harder tests, I don't know... luck I guess...</li>
<li>23 people passed Test 30 (first bonus test), but only two people (I.E. one module) passed Tests 31 or 32. This means that 21 people had code which was <i>capable</i> of handling really complex networks with more than 2 trains, but were let down by a bug somewhere in their code (which most likely would've caused failure in some of the easier cases too).</li>
</ul>
<p>(samc)</p>



[an error occurred while processing this directive]