Unified Diff Viewer
One column, with removed lines and added lines interleaved in place. This is the format git prints and code review tools display, so it is the one to use when the output is going into a commit message, a ticket or a chat message.
This page opens in unified view.
Reading it
A removed line is followed immediately by its replacement, so a small edit reads as a tight pair rather than requiring you to look across two panes. Unchanged context lines sit between the changes to anchor them.
It stays readable at narrow widths, which makes it the better choice on a laptop or in a split editor.
- timeout: 30
+ timeout: 60
retries: 3Moved blocks look like churn
A standard diff has no concept of movement: a block that relocated is absent from one position and present at another, which is precisely a deletion plus an insertion. That is the algorithm being correct rather than unhelpful, but it does mean a large reorganisation produces a diff that overstates how much really changed.
Normalise before comparing structured data
A textual diff of two JSON documents reports reordered keys and different indentation as changes even when the parsed value is identical. Format both sides first and the diff collapses to what actually differs.
Context lines and patch format
Unified output with surrounding context is the shape of a patch file, which is why it can be pasted into a review or an issue and read by anyone familiar with git.
It is not a valid patch without file headers and hunk ranges, so do not expect the output to apply with git apply - it is for humans reading a change, not for tooling.