Linear Sampling Algorithm

The LSA is a sampling-based strategy created by Flávio Mota Medeiros to detect configuration-related bugs in Software Product Lines. The strategy groups three sampling methods (one enabled, one disabled, most enabled disabled) to generate the configurations that need to be tested.

Running

The algorithm can be executed using one of the following instructions:

Generate model

python main.py -p project_name -g project_path

Parameters:

• -p: name of the project, used to name the model and features files;
• -g: path of the project.

Model Example:

{
    "1": {
        "type": "Feature",
        "attr": "optional",
        "name": "A"
    },
    "2": {
        "type": "Feature",
        "attr": "optional",
        "name": "B"
    }
}

Run LSA

python main.py -p project_name -i model.json -t type

Parameters:

• -p: name of the project, used to name the result file;
• -i: path of the model.
• -t: type of strategy, can be:
  • original: Generate results basying only on features;
  • any: Generate results respecting the limitations of the model (only one product is created in decision making);
  • all: Generate results respecting the limitations of the model (all products are created in decision making);
  • rand_original: Same as the original type, but only 3 products are created;
  • rand_any: Same as the any type, but only 3 products are created;
  • rand_all: Same as the all type, but only 3 products are created.

Authors

• Gleyberson Andrade - gleybersonandrade

References

• An Approach to Safely Evolve Preprocessor-Based C Program Families (MEDEIROS; Flavio, 2016)

License

This project is licensed under the MIT License.