What Is a Diff? Understanding File and Text Differences

The word "diff" comes from the Unix diff utility, first released in 1974. Its job was simple: given two text files, output the minimum set of changes needed to transform one into the other. That output — the diff — became the foundation for version control systems, code review workflows, and collaborative editing as we know them today.

How a Diff Is Computed

At its core, a diff answers one question: what is the shortest sequence of additions and deletions that converts file A into file B? This is known as the edit distance problem, and it turns out to be computationally non-trivial for large files. The breakthrough came in 1986 when Eugene Myers published an algorithm that finds the shortest edit script in O(ND) time, where N is the sum of both files' lengths and D is the number of differences. The Myers algorithm is now the industry standard, used in Git, GNU diff, and most modern comparison tools.

LineDiff extends the Myers algorithm with semantic cleanup. After computing the raw edit script, the engine refines it to produce diffs that align with human-readable boundaries — word edges, sentence breaks, meaningful structural units. This means the highlighted changes are the ones a human would naturally point to, not artifacts of the algorithm's internal operations.

Granularity: Line, Word, and Character

Diffs can operate at different granularities. Line-level diffing is the most common: each line is treated as an atomic unit, and the diff shows which lines were added or removed. This works well for code and structured data where lines carry distinct meaning. Word-level and character-level diffing goes further, highlighting the specific words or characters that changed within a line. LineDiff performs both simultaneously — line-level for the overall structure, then word and character inline diffs for the specific changes within each modified line.

Unified View vs Split View

There are two primary ways to display a diff. Unified view, also called inline view, presents the two versions as a single stream. Deleted lines appear in one color, added lines in another, and unchanged context lines sit between them. This is the format produced by git diff in the terminal. Split view, also called side-by-side view, shows the two versions in adjacent panels, with corresponding lines aligned horizontally. Split view is generally easier for humans to read when the files have significant structural differences, because you can visually track how a passage transformed.

A Practical Example

A practical example makes the difference clear. Suppose you have two versions of a project README file. The unified view shows a block of deletions followed by a block of additions — you have to mentally reconstruct what the paragraph looked like before and after. The split view places the old paragraph on the left and the new paragraph on the right, with the specific changed words highlighted. Your brain processes this comparison much faster.

Online Tool, Desktop App, or Command Line?

The question of when to use an online tool versus a desktop application or command-line utility comes down to access and collaboration. Desktop tools like Beyond Compare or Araxis Merge are powerful but require installation and are licensed per machine. Git diff is always available in a repository but produces terminal output that is difficult to share or annotate. Online tools like LineDiff require no installation, work from any device, and make sharing a comparison as easy as sending a link. They also support file formats beyond plain text — PDF, Word, Excel, JSON, YAML, and more — that command-line diff tools cannot handle natively.

For most people comparing documents, configurations, or code snippets outside of an active Git workflow, an online diff tool offers the fastest path from two files to a clear, shareable visualization of what changed.