Soort van start met hfdst5

diff --git a/a3/5discussion.tex b/a3/5discussion.tex
index 4faa6f2..87a78c1 100644 (file)
--- a/a3/5discussion.tex
+++ b/a3/5discussion.tex
@@ -8,4 +8,28 @@
 % • Which method of test test selection is used?
 % • What is the modeling notation, its expressiveness, and ease of use? 
 % • Do the tools use on/off-line testing (on-the-fly/batch)?
-% • Do they support both: test input generation and output checking?
\ No newline at end of file
+% • Do they support both: test input generation and output checking?
+
+The SUT has been tested using two tools, JTorX and \GAST. Both tools allowed
+for extensive testing of the SUT compared to the model of 
+Section~\ref{sec:modeling} using the architecture of Figure~\ref{fig:arch}.
+
+However setting up both \GAST and JTorX required extensive effort. To be able 
+to use \GAST extensive changes had to be made to the source code of \GAST. 
+Without these changes \GAST could not be used effectively. This was due in 
+part to problems with Cleans uniqueness typing when using \GAST over TCP. The
+complete list of changes can be obtained by comparing \GAST s source code to the
+source code found in the attached \texttt{code/Gast/} folder.
+
+\GAST selects tests by trying all inputs from the alphabet in each possible 
+state. When the SUT responds in a way which is not in the model the test case
+fails. JTorX selects test cases by ???????????????
+\GAST s selection ensures that all possible traces are covered, however, this
+exhaustive search will also generate huge tests for big models. 
+
+% hier iets over non-determinism?
+
+Both JTorX and \GAST use on-line checking. They test the SUT on the fly by
+syncing state transitions in the internal model of the SUT with observed output (after input) from the SUT.
+
+%iets over ease of use van test model generation
\ No newline at end of file