Tailwind Config Report

Some mildly interesting details from analyzing 2788 open source projects that rely on Tailwind CSS. I am not affiliated with Tailwind Labs, just a casual user of their projects. Originally, this post got me curious about some of the practices, so I hacked together some past scripts and similar experiments to generate these results (source code). I have not spent too much time reviewing data quality, so there is still room for improvement.

The following summaries use "Source Owners" to refer to unique organizations/users owning a repository and "Source" to refer to a Tailwind-related path within a repository. Some owners may have multiple repositories, and some repositories may have multiple sources. Gauging by source owners seemed like a slightly better indicator of usage. If only a single source owner makes a particular configuration change, it is generally excluded from the summary tables.

• General Usage - customized features, disabled core plugins, content file extensions, prefixes
• Analyzed Versions - related package versions which were resolved and used during analysis
• Version Constraints - related package version constraints documented in package.json files
• All Packages - all packages referenced in package.json files
• CSS Variable Names - common variable names found
• Config File Duplication - duplicate configs used across sources
• Theme Properties - animation, aspectRatio, backgroundImage, backgroundPosition, backgroundSize, blur, borderRadius, borderWidth, boxShadow, brightness, color, columns, content, contrast, cursor, flex, flexGrow, flexShrink, fontFamily, fontWeight, gridAutoRows, gridColumn, gridColumnEnd, gridColumnStart, gridRow, gridRowEnd, gridRowStart, gridTemplateColumns, gridTemplateRows, hueRotate, letterSpacing, lineHeight, listStyleType, maxWidth, minHeight, minWidth, opacity, outlineOffset, outlineWidth, ringWidth, rotate, saturate, scale, skew, spacing, strokeWidth, textDecorationThickness, textUnderlineOffset, transformOrigin, transitionDelay, transitionDuration, transitionProperty, transitionTimingFunction, willChange, zIndex
  • Updated Names - shows built-in names where users overwrite default values
  • Additional Names - shows user-introduced names and their values
  • Value Conventions - shows basic usage of units (e.g. rem, px), CSS functions, and standard keywords (only supported for some properties)
  • Deleted Names - shows built-in names where users disabled them

Data

In total, 3081 analysis tasks were executed (2788 completed successfully, 293 failed; 2609 used dependency squashing, 472 full installation). Tasks can fail for a number of reasons including: wide range of expected Node build runtimes, outdated package managers, restricted network access and build runtime, and occasional bugs.

The following raw data files are included in this branch for each task:

• extract-completed (or extract-failed) is an empty marker file which indicates the job completed successfully
• extract-installed-packages.jsonl contains the packages and versions used during the analysis (which may be a subset of the original package.json file)
• extract-metadata.jsonl contains some metadata about the analysis such as project source, times, and analysis behaviors
• extract-raw-package.json contains the original contents of the package.json file from the source
• extract-raw-tailwind-config.js contains the original contents of the tailwind.config.* file from the source
• extract-tailwind-config-baseline.json contains the baseline tailwind configuration with common, default plugins and no customization
• extract-tailwind-config-effective.json contains the effective tailwind configuration with any customizations
• extract.log contains the console output generated during analysis

These public sources were discovered primarily via GitHub API based on code search and tailwind-tagged repositories. Some additional requirements were applied, too, such as semi-recent commits, stars, and not being forks. The following are included in this dataset, although this is a small subset of the potential sources which could be analyzed.