#Conflerge Conflerge is custom mergetool that strives to automatically resolve merge conflicts. This is an attempt to demonstrate that finer grain merging, unlike the line-by-line merging in Git, can result in fewer false-positive conflicts with a competitive false-negative rate.

We attempted finer grain merging with two approaches: Token-based and Abstract Syntax Tree based.

###Tokenization A token is an individual keyword in source code such as a reserved keyword, variable name or constant. For example, tokenizing the Java statement System.out.println(hello); yields the token sequence [ System . out . println ( hello ) ; ] We use the Wagner-Fischer Algorithm [1] to find the edit distance between BASE and LOCAL, and between BASE and REMOTE.

###Abstract Syntax Trees In contrast to tokens, Abstract Syntax Trees retain information about the syntactic structure of the code as hierarchical parent-child relationships. Each node has an ordered set of children. We use the mmdiff algorithm [2] to find the edit distance between BASE and LOCAL, and between BASE and REMOTE.

#Architecture

Conflerge is implemented as a mergetool. Git provides the relevant versions of BASE, LOCAL and REMOTE. We use javaparser to parse and unparse the source code files. We also use it for tokenization and the construction of abstract syntax trees.

#Usage

Conflerge is currently scoped to work with Java source code files and Git.

To use the tool:

• Clone the Conflerge repository to ~/.
• Update your .gitconfig to include:

    [mergetool "conflerge-tree"]
		cmd = java -jar ~/Conflerge/ConflergeTree.jar \$BASE \$LOCAL \$REMOTE \$MERGED
    [mergetool "conflerge-token"]
        cmd = java -jar ~/Conflerge/ConflergeToken.jar \$BASE \$LOCAL \$REMOTE \$MERGED
    [merge]
        tool = conflerge-tree
        tool = conflerge-token

• To run with token-based merging: git mergetool --tool=conflerge-token <conflicting file>

• To run with AST-based merging: git mergetool --tool=conflerge-tree <conflicting file>

• Note: Our experiments showed that AST-based merging was the more successful strategy, so we recommend using conflerge-tree.

Instructions to recreate the results in the paper can be found here.

###References [1] Gonzalo Navarro. 2001. A guided tour to approximate string matching. ACM Comput. Surv. 33, 1(March 2001), 31-88.

[2] Sudarshan S. Chawathe. 1999. Comparing Hierarchical Data in External Memory. In Proceedings ofthe 25th International Conference on Very Large Data Bases (VLDB ’99), Malcolm P. Atkinson, Maria E.Orlowska, Patrick Valduriez, Stanley B. Zdonik, and Michael L. Brodie (Eds.). Morgan Kaufmann PublishersInc., San Francisco, CA, USA, 90-101