Artifact

Artifact package for our paper “How do Developers Talk about GHA?”. This repository includes our data and scripts.

Data

1. Data collection (2018.10.1-2022.10.31)
   • SO data, i.e., posts, from the official SO data dump
   • GitHub data, i.e., issues, using the GitHub Search API
2. Data for manual classification: SO post and GitHub issue
   • This data includes:
     • 6,590 SO questions (Q_S) with 2,471 accepted SO answers (A_S)
     • 315 GitHub issues (Q_G) with 217 closed GitHub issues (A_G)
   • The results of manual classification can be found in all_post_issue_category.csv
   • Data structure: (id, type, phase, category)
     • id: the number used in this paper. “P1” and “I1” represent the first SO post and the first GitHub issue in our dataset, respectively.
     • type: “github issue” or “so post”
     • phase: phase of a post or an issue
     • category: category of a post or an issue
3. Data for characteristics analysis
   • The data for characteristics analysis can be found in so_post_popularity.csv and so_post_difficulty.csv
   • Popularity metrics include:
     • avgView, the average number of views for all the questions of a category;
     • avgFav, the average number of favorites for all the questions of a category;
     • avgScore, the average score for all the questions of a categpru;
     • avgAns, the average number of answers for all the questions of a category.
   • Difficulty metrics include:

     - ansRate, the percentage of questions of a category with at least one answer;
     - acceptRate, the percentage of questions of a category that have accepted answers;
     - timeFA, the median time needed for questions of a category to receive the first answers, in hours;
     - timeAA, the median time needed for questions of a category to receive the accepted answers, in hours;
     - textSize, the average number of description characters for questions of a category.

4. The accepted answer examples and detailed discussion of each solution strategy can be found in solution_strategies.md

Script

We seek to analyze the characteristics of the identified problem categories in terms of popularity and difficulty.

• Spearman’s rank correlation coefficient cor.R

Figure

• Figure 1: The trend of GHA discussed on Stack Overflow
• Figure 2: The taxonomy of GHA problems

Table

• Table 1: Popularity of GHA problem categories
• Table 2: Difficulty of GHA problem categories
• Table 3: Correlation between Popularity and Difficulty of GHA problem categories
• Table 4: Difficulty of GHA problem categories (GitHub issues)