Pink pony is an open-source software command-line application that suggests functional clusters based on the common changes on git. Functionally related files are grouped into the same cluster. In this way the tool can help to split a monolithic application into a set of microservices.
What changes together should be also together
The application produces 2 files:
cluster.txt
: contains clusters;graphViz.dot
: a .dot file to visualize the co-change graph.
To run the Pink pony application you have to download the recently released version.
mr
: Markov Clustering is a hard clustering algorithm;ch
: Chinese Whispers is a hard clustering algorithm;max
: MaxMax is a soft clustering algorithm for undirected graphs;watset
: Watset is a local-global meta-algorithm for fuzzy graph clustering.
The implementation of the algorithms used from Watset project
Depending on the investigation that you want to do, the application provides two options:
-
file
: Suggests functional clusters on file level. -
module
: Suggests functional clusters on module level. -
This will suggest functional clusters from files.
$ java -jar pinkpony.jar path\to\.git file max
- This will suggest modules under the specific module
moduleName file1 file2 ... fileN
cluster1 file1 file3 cluster2 file6 file89 ... clusterN file2 file15
$ java -jar pinkpony.jar path\to\.git file max moduleName
- The following command will suggest new co-change modules from the submodules of mod1, mod2, mod3 with the Markov Clustering algorithm.
mod1 subMod1 sudMod2 subMod3 mod2 subMod4 sudMod5 mod3 subMod6 sudMod7 subMod8 subMod9
cluster1 subMod3 subMod9 subMod8 cluster2 subMod4 subMod5 subMod1 ...
$ java -jar pinkpony.jar path\to\.git module mr mod1 mod1 mod2
The Pink Pony application is a part of my thesis project.