- \n
- Promote\u00a0data integration with other existing development support\u00a0tools.<\/li>\n
- Enable to write\u00a0queries on\u00a0Git data using standard SQL syntax.<\/li>\n<\/ul>\n
To ensure efficiency, Gitana\u00a0relies on\u00a0an incremental propagation mechanism that refreshes the database content\u00a0with the latest modifications in Git repositories.<\/p>\n
Full details on the Git conceptual schema we propose are available in our ER’15<\/a>\u00a0accepted paper that you can read here<\/a>. Read on for a summary of the work.<\/p>\n
Approach<\/h2>\n
Gitana<\/a> proposes a conceptual schema for Git<\/strong> to facilitate data integration between existing\u00a0Git-based tools and advanced query operations. This schema is materialized as a\u00a0relational database, for which we have defined an extraction process that populates\u00a0and keeps it up-to-date and an export process to perform analysis in other technologies.<\/p>\n
Conceptual model for Git data<\/h2>\n
The structure of a Git repository<\/em> is shown in the conceptual schema of Fig.1. Note that the\u00a0conceptual schema also includes some derived concepts, methods and attributes to facilitate\u00a0the analysis of Git repositories by making explicit some information that is\u00a0normally hidden or hard to query. This is the case for LineDetail<\/em>\u00a0( that stores information for each line of textual files) and FileRename <\/em>\u00a0(that keeps tracks file rename actions), the getVersion<\/em> method\u00a0(used to get\u00a0the version of a file at a particular timestamp)\u00a0and the attributes filesDeleted<\/em> and emptyLines <\/em>(used to provide some\u00a0activity metrics).<\/p>\n
![\"conceptualModel-ext-B\"](\"https:\/\/modeling-languages.com\/wp-content\/uploads\/2015\/08\/conceptualModel-ext-B-1024x635.png\")
<\/a>Fig. 1 – Conceptual model.<\/p><\/div>\n
Database schema for Git data<\/h2>\n
The previous conceptual schema is materialized in the relational database schema of\u00a0Fig. 2. In a nutshell, concepts\/attributes in the conceptual schema are mapped into tables\/columns in the database schema and associations are mapped into foreign keys\u00a0or new tables\u00a0depending on the cardinality of the association, following\u00a0the typical translation strategies.
\nAdditionally, the derived attributes in the conceptual schema are mapped as views, while\u00a0auxiliary methods are\u00a0implemented as either functions or stored procedures.<\/p>\n![\"schema-v5\"](\"https:\/\/modeling-languages.com\/wp-content\/uploads\/2015\/08\/schema-v5.png\")
<\/a>Fig. 2 – Database schema for Git.<\/p><\/div>\n
Integration with other development tools<\/h2>\n
The schema presented in Fig. 2\u00a0can be integrated with other development-related data\u00a0coming from tools that rely on a database infrastructure. Examples of such tools are\u00a0most of issue tracking systems (e.g., BugZilla, Trac, Mantis) plus tools like GHTorrent<\/a>, a scalable and offline mirror of GitHub; Gerrie<\/a>, a data and information\u00a0crawler for Gerrit, and Bicho, a tool that is able to parse different issue tracking\u00a0systems (e.g., Launchpad, Jira, Allura).<\/p>\n
Any of such tools embeds at least one concept that deals with <\/span>Commits<\/span><\/i>, <\/span>Files<\/span><\/i> or <\/span>Developers<\/span><\/i>. By leveraging on such concepts we can connect their database schemas with our Git schema by matching file names and SHA identifiers for Commits<\/em> and Files<\/em>, and relying on\u00a0identity matching algorithm\u00a0for<\/span>\u00a0Developers.<\/em><\/p>\n
As an example, we show\u00a0in Fig. 3\u00a0how our Git schema\u00a0can be integrated with GHTorrent<\/a> and\u00a0Gerrie<\/a> databases. The integration\u00a0with GHTorrent provides a broader view of a GitHub project by combining\u00a0the collaboration data and the Git SCM data, which GHTorrent does not cover at the\u00a0moment. On the other hand, Gerrie integration allows extending the analysis of\u00a0code-reviews with fine-grained information from the Git SCM data.<\/p>\n
![\"ghTorrentAndGerrieAndGitana_v2\"](\"https:\/\/modeling-languages.com\/wp-content\/uploads\/2015\/08\/ghTorrentAndGerrieAndGitana_v2.png\")
<\/a>Fig. 3\u00a0– Integration scenario.<\/p><\/div>\n
Advanced query functionalities on Git data<\/h2>\n
Given the database representation of the SCM system, we can leverage on the plethora\u00a0of tools and techniques existing in the database realm to perform advanced queries on\u00a0the data. In\u00a0the example below we show how to count the modifications (i.e., additions, deletions and changes) made by each developer in the repository. As can be seen,\u00a0instead of using a combination of Git and Shell commands, with Gitana, pure\u00a0SQL syntax suffices to get the information.<\/p>\n
![\"example\"](\"https:\/\/modeling-languages.com\/wp-content\/uploads\/2015\/08\/example-1024x284.png\")
<\/a>![\"example-2\"](\"https:\/\/modeling-languages.com\/wp-content\/uploads\/2015\/08\/example-2-300x143.png\")
<\/a>