Analytical Questions Based on Debsources Paper

[SahithiKasim]

Upstream code analysis:

• 1 a. What trends can be observed in code complexity over time-related to the?

• Lines of Code (LOC): Count the total lines of code in a codebase. Higher LOC generally indicates greater complexity.

• Dependency Counts: The number of dependencies of packages. More dependencies implies greater complexity.

• Code Churn: The number of lines of code added/deleted over time. High churn suggests complex code that requires more frequent changes.

• 1b. How has the cyclomatic complexity of core Debian packages changed across different time periods?

• Get the source code history for each package from its git repository

• Calculating the cyclomatic complexity for each package version over time using static analysis tools

• Apply statistical analysis (like regression) to test for a significant trend in mean cyclomatic complexity

• How has the diversity of licenses used in the codebase evolved?

• Extract license information from Upstream copyright files.

• Map the license identifiers to general categories like BSD, GNU, Apache, OpenSSL, etc. (can be found in this https://www.debian.org/legal/licenses/)

• For each year, tally the number of packages under each license category. Calculate category proportions.

• Analyze the trends in license category proportions over time using descriptive statistics.

• What changes have occurred in the usage of programming languages?

• Examine source code files and identify programming languages using heuristics, file extensions, and tools like cloc.

• Categorize and tally usage for languages like Python, Java, C/C++, Rust, etc.

• Track the proportions of each language over time as packages evolve.

• Apply statistical analysis to determine if language proportions have significantly changed.

• What is the correlation between authors and licenses, and how have contributions varied over time?

• For each package, analyze git commit history.

• Extract the author's name and emails for each commit and match the author's name with the maintainers table. Resolve aliases to unique contributors.

• Aggregate commits by contributor to determine top contributors and their activity levels.

• Analyze email domains to categorize contributors by organization (e.g. @debian.org, @redhat.com). Calculate organization proportions over time.

• Categorize licenses.

• Use correlation analysis to identify relationships between authors/organizations and license categories.

[SahithiKasim]

Preliminary findings from 500 cloned GitHub repositories include metrics - lines of code (LOC), code churns, dependency counts, and the languages used.

250 repo results can be found here - results.md
I am updating this sheet regularly so you can find latest results in this - https://docs.google.com/spreadsheets/d/1UktUekTQd__dGa-vd--p27k6zoUDcPo6cft-jRk3dJ8/edit?usp=sharing