One Wise Sampling

The One Wise Sampling plugin allows one wise sampling for cardinality-based feature models. It generates a non-minimal set of random valid configurations that cover the model one wise under boundary interior coverage and instance set coverage and outputs them into the console.

The One Wise Sampling plugin requires the model to be bound which means no infinite upper bounds as instance cardinalities are allowed. In case of an unbound model, you can use other plugins like the Big M plugin to replace infinte upper bounds with finite ones.

Usage

Import a cfm and generate a one wise sample set for it:

python3 -m cfmtoolbox --import example.uvl one-wise-sampling

In case of an unbound model, transform the model first with e.g. the Big M plugin.

python3 -m cfmtoolbox --import unbound.uvl --export bound.uvl apply-big-m
python3 -m cfmtoolbox --import bound.uvl one-wise-sampling 

Because the sampling algorithm uses non-determinism, it is recommended to limit the runtime of the command with a timeout of e.g. 5 seconds.

timeout 5 python3 -m cfmtoolbox --import example.uvl one-wise-sampling

To store the sampling in a .json file, shell redirection can be used, as shown in the following example:

python3 -m cfmtoolbox --import example.uvl one-wise-sampling > sampling.json