Glossary¶

This glossary defines the terms Codefang introduces or uses in a specific way. Terms are listed alphabetically.

Abstract syntax tree (AST)¶

A tree representation of source code that captures its structure rather than its raw text. Codefang analyzes the AST so it can reason about meaning, not just characters.

Analyzer¶

A pluggable unit of analysis. Each analyzer carries an ID (for example static/complexity or history/burndown) and produces metrics. Static analyzers consume parsed code; history analyzers consume commit history.

Bare repository¶

A Git repository without a working tree. Codefang's history mode opens both normal and bare repositories via libgit2.

Blob¶

The raw content of a file at a specific Git object hash. History analysis reads blobs to compute diffs and parse code.

Burndown¶

A history analysis that tracks code age and survival: how many lines written in a given period are still present in the codebase over time. The result is often shown as a burndown chart.

Change filter¶

The per-run rule that decides which changed files a history analyzer sees: --languages, the vendored and generated path policy (--include-vendored, --include-generated, --extra-excluded-prefixes), the generated-content marker in the first 512 bytes, and --max-changes-per-commit.

Chronological reduce¶

The single-threaded, oldest-first fold of every commit's product into an analyzer's report. Doing it in one fixed order is what makes history output byte-identical across runs and machines.

Cohesion¶

A static metric measuring how strongly the functions of a file belong together, computed as LCOM over the variables each function reads or writes. Codefang reports 1 − LCOM as the cohesion score.

Commit window¶

The commits a history run analyzes: the newest --limit commits, or every commit authored since --since, or just HEAD with --head, delivered to the analyzers oldest-first.

Couples¶

A history analysis that detects co-change coupling: files that tend to change together in the same commits. Useful for risk-aware refactoring.

Halstead¶

A family of static software-science metrics (volume, difficulty, effort, and related measures) derived from the operators and operands in code.

History mode¶

The analysis mode that walks a Git repository's commit history and runs history analyzers. Contrast with static mode, which reads source files from disk.

Identity detection¶

The step (IdentityDetector) that maps commit authors to canonical developer ids, so one developer with several names or emails counts as one person. By default a known email or name unifies the two; --exact-signatures and --people-dict change the rule.

Per-commit product¶

The order-independent work for one commit (tree diff with rename detection, line diffs, blob reads, UAST parses) computed in parallel before the chronological reduce. Its result depends only on that commit.

Sentiment¶

A history analysis that scores the sentiment of code comments over time.

Shotness¶

A history analysis that measures function-level change frequency — which functions are touched most often (structural hotness).

Static mode¶

The analysis mode that reads source files from disk, parses them into UASTs, and runs static analyzers. It requires no Git history.

Tick¶

A discrete time index assigned to commits so analyses can build time series. The first commit's committer time, floored to the tick size (--tick-size, default 24 hours), is tick 0; ticks never go backwards.

Typos¶

A history analysis that detects likely identifier typos in diffs.

UAST (Universal Abstract Syntax Tree)¶

A language-neutral abstract syntax tree. Codefang parses 60+ languages into a single UAST shape via Tree-sitter, so one analyzer (for example complexity) works across every supported language.