Say I have a test set - e.g. Registration, this contains all tests associated with registering a patient.
If a test fails, has a defect raised, which is subsequently fixed - how should I re-run that test.
Options (as far as I can tell) are;
1. Simply re-run the same test instance
2. Create a new test instance in the same Test Set
3. Create a new Test Set for fixed defects
It all depends on the reporting you have to provide.
Usually, a separate reporting is done for each "build" of the System Under Test (a "build" is a version of the SUT you're given to test). In that case, solution 3 is probably more appropriate (besides the fact that it can also account for some regression testing, i.e. the new test sets would include not only the one that failed but also some that passed).
Solution 1 is good enough if you don't care to report for each build, but just for the complete SUT.
Solution 2 is kind of in-between that you can use if you are not sure about the reporting, although it is probably the one that will give you the more headaches to build the filters.
In Test Lab, we assign a cycle value to each test in a test set for a given release. If we have a test fail in our first cycle we would pull over another instance of the test and assign it to cycle 2 in the same test set. This way you can filter on the cycle value and report only the results for that cycle while still using the same test set.
If you go with Option 2, you will be adding more test cases to the execution list. Which may not have been planned in the original test plan. Not sure if that may cause any mis-conceptions in your reporting.
Option 1: I would say is the best option. This way you can filter the same test set for the kind of report that you want to present.
Say you have a test set with 30 test cases with 10 failed ones.
With a new build you want to re-test and report on these 10 failed test cases.
You can filter to show all fail cases, execute them and now you can filter the set based on time or date and get the last 10 executed test cases and report on them.
It's always handy to re-use the test case, because there is a history of each and every run. Yes its a bit of pain with QC 9.X in the reporting aspect, but its a trade off.