Difference between revisions of "Slicer-3.6-QA"

From Slicer Wiki
Jump to: navigation, search
Line 22: Line 22:
  
 
The '''data processing''' section can be tested by using standard CTest/CMake mechanisms. Basically by adding ADD_TEST() entries to the CMakeLists.txt file of the module.
 
The '''data processing''' section can be tested by using standard CTest/CMake mechanisms. Basically by adding ADD_TEST() entries to the CMakeLists.txt file of the module.
 
  
 
=Luis Ibanez' scoring system=
 
=Luis Ibanez' scoring system=

Revision as of 14:52, 15 April 2010

Home < Slicer-3.6-QA

Return to Slicer 3.6 documentation

  • This page contains our assessment of the Slicer 3.6 modules and extensions
Slicer 3.6 Quality Assurance
Score Name Documentation Help [1] Acknowledgment [2] Test coverage [3] valgrind errors
Gold my module complete yes and yes yes, yes, yes 80% 0
  1. including link to website documentation
  2. including name of author and of supervisor
  3. in % for non-interactive part

Testing Partition

Most Slicer modules have a GUI component and a Data Processing component.

Testing GUI components is still a challenge, so we will focus here on testing the Data Processing components. This can be done in most cases by partitioning the module into a GUI section and a Data Processing section, where the second one usually takes the form of a C++ class (although that is not a requirement).

The data processing section can be tested by using standard CTest/CMake mechanisms. Basically by adding ADD_TEST() entries to the CMakeLists.txt file of the module.

Luis Ibanez' scoring system

The following scoring will be applied to the data processing sections of all modules:

Slicer 3.6 Quality Assurance Scoring
Score Code Coverage Valgrin Errors Documentation Tutorial
Gold > 80% 0 yes yes
Silver > 70% < 10 yes yes
Bronze > 60% < 50 yes yes
Aluminium > 50% < 100 yes yes
Coal > 50% > 100 yes yes
Plutonium unknown unknown no no

The code coverage and Valgrind error must be the ones reported on the Nightly Slicer Dashboard. Anecdotal data is not acceptable.