|
|
Download
| User
Forum |
Gallery Contact Us |
| Home > Collaboration > Working Groups > System Testing Team |
System Testing Team
The System Test Team was established at the Geant4 Niigata Workshop in July 1998. The Working Group is currently composed by:
STT AimsTo improve the quality and robustness of GEANT4 code at the "System Level", that is to say, to ensure it compiles and runs without crashing on all supported platforms in tests wich cover a wide range of physics, geometries, particles, processes over many events. Although physics validation is not our aim, we will obviously be aware of the physics quality and seek cooperation with the category teams.STT Assumptions and RequirementsDuring normal development, tags will be tested and simply accepted or rejected - we call this incremental acceptance. Two weeks before a public release we go into the release phase - see below.
Communicating "Incidents" to Category CoordinatorsFor each Test Incident, i.e., for each problem or bug, the STT member should make a Test Incident Report as follows:
Test SuiteA set of tests in geant4/tests/ excercises the code. Also many examples are used for regular integration testing.Test ProceduresThere are detailed test procedures for members of the System Test Team.Tags DatabaseCategory tags are notified and recorded in a database, tested then accepted or rejected according to the above procedures.Release PolicyHopefully, the above policies will mean that we go into the release phase with no outstanding problems. However, some new problems might emerge during the porting of the code on all supported platforms/compilers. There might also be some agreed "deferred" problems. If these cannot be fixed, and are not killers, they should be listed in the Release Notes. As also specified in our Tag and Release Policy, if required to create a CVS branch, Testers - including Category Testers - will have to checkout the branch into a separate working directory so as to avoid confusion with the main trunk (development) version. The branch will then be used for testing and any bugs found will have to be fixed there and, if appropriate, also in the development version. This will be done primarily by the Category Testers in response to TIRs. A CVS branch behaves like an independent version of the code. This means that developers are free to continue developing. The down side is that System and Category Testers have to check out an independent version of the code in a separate working area so that essential bugs (ESSENTIAL bugs only) can be fixed on the branch (and also be fixed in the HEAD if appropriate). But we have discussed this and we think it is an acceptable price to pay for giving us the freedom to patch releases and continue development.Work plans |
Last updated: 05/20/2010