Where is an implementation for Simple Recovery Algorithm?

[ty-bnn]

I am making changes to GumTree Algorithm for my research.
When I run webdiff with no options except file paths, what kind of recovery algorithm are applied by default?
According to your paper, Fine-grained, Accurate, and Scalable Source Code Differencing, you have three kind of recovery algorithms, which are optimal, simple and hybrid of them.
And I want to know where the Simple Recovery Algorithm is implemented?

[jrfaller]

Hi @ty-bnn !

The Simple algorithm is the default one starting from GumTree 4. You can take a look at the implementation of most GumTree's matchers starting in this class : https://github.com/GumTreeDiff/gumtree/blob/main/core/src/main/java/com/github/gumtreediff/matchers/CompositeMatchers.java.

Also, don't hesitate to take a look into our auto matchers that are based upon the work we did on hyper parameter optimization here : https://github.com/GumTreeDiff/gumtree/blob/main/core/src/main/java/com/github/gumtreediff/matchers/AutoMatchers.java.

Cheers!

[ty-bnn]

Hi @jrfaller
Thanks for sharing tips!
When I run GumTree 4, it seems to work SimpleGumTree as a matcher as described here, which means the tool runs GreedyMatcher as Top-Down phase and SimpleBottomUpMathcer as Bottom-Up and Recovery phase right?

[jrfaller]

Exact!

[ty-bnn]

I understood.
Thank you!
Have a happy new year!!