Difference between revisions of "2013 Summer Project Week:Sample Data"
From NAMIC Wiki
Line 43: | Line 43: | ||
** Good idea and technically feasible, but can be independent of the sample data issue | ** Good idea and technically feasible, but can be independent of the sample data issue | ||
* The plan for now | * The plan for now | ||
− | ** Add an application-level (qSlicerApplication) container for Sample Data descriptions (name, URL) | + | ** Add an application-level (qSlicerApplication) container for Sample Data descriptions (name, URL, md5) |
** Scripted/Loadable modules can register their sample data resources | ** Scripted/Loadable modules can register their sample data resources | ||
** CLI factory can be augmented in the future to populate this container based on CLI tags | ** CLI factory can be augmented in the future to populate this container based on CLI tags | ||
+ | ** Add application functionality to manage cache to minimize re-download | ||
Latest revision as of 14:18, 21 June 2013
Home < 2013 Summer Project Week:Sample DataKey Investigators
- Isomics: Steve Pieper
- GE: Jim Miller
- Kitware: Jean-Christophe Fillion-Robin
- Noware: Bill Lorensen
Objective
Develop a plan so that modules (and extensions) can define sample data.
Approach, Plan
- Build from ExternalData as currently used in CTest
- Use URLs and hashes to identify the data
- Support local caching to minimize downloads
- Integrate into SampleData module GUI
- Also consider related topic of CLI modules defining self-tests
- refer to sample data, parameters, and baseline results
- integrate these automatically into ctest and the SelfTests module
Progress
- Reviewed and discussed the current CMake/CTest implementation of hash-based data access
- Cannot reuse CMake implementation
- Some features like git hooks not really needed
- Ideally re-use the cache and naming conventions as much as possible
- Reviewed ideas for adding testing and sample data tags to XML
- Good idea and technically feasible, but can be independent of the sample data issue
- The plan for now
- Add an application-level (qSlicerApplication) container for Sample Data descriptions (name, URL, md5)
- Scripted/Loadable modules can register their sample data resources
- CLI factory can be augmented in the future to populate this container based on CLI tags
- Add application functionality to manage cache to minimize re-download
Delivery Mechanism
This work will be delivered to the NA-MIC Kit as a Slicer module, ultimately integrated into the slicer core