[gghez]

Summary

rtk diff incorrectly reports "Files are identical" when lines differ by only a few characters. Two root causes:

1. The identical-files guard at line 24 only checked added == 0 && removed == 0, ignoring the modified counter. Lines classified as modifications (similarity > 0.5) incremented modified but not added/removed, so the guard passed and the diff silently swallowed actual differences.

2. The similarity() function used Jaccard similarity on unique character sets, which over-estimated similarity for lines sharing common characters regardless of position. Replaced it with a positional prefix+suffix ratio that better reflects actual line differences.

Changes

• Added && diff.modified == 0 to the identical-files guard in run()
• Replaced similarity() implementation: Jaccard set-based -> positional prefix+suffix metric
• Added 4 regression tests covering all reproduction cases from the issue
• Updated existing test_similarity_partial_overlap for the new metric

Test plan

• All 983 existing tests pass (cargo test --all)
• 4 new regression tests cover exact reproduction cases from diff reports "Files are identical" when lines differ by only a few characters #781
• cargo check compiles without warnings

Fixes #781

[CLAassistant]

All committers have signed the CLA.

[pszymkowiak]

[w] wshm · Automated triage by AI

📊 Automated PR Analysis

🐛 Type	bug-fix
🟡 Risk	medium

Summary

Fixes a bug where rtk diff falsely reported files as identical when lines differed by only a few characters. The identical-files guard now checks the modified counter in addition to added and removed, and the similarity function was replaced with a positional prefix+suffix metric to avoid over-estimating similarity.

Review Checklist

• Tests present
• Breaking change
• Docs updated

Linked issues: #781

---

Analyzed automatically by wshm · This is an automated analysis, not a human review.

[aeppling]

Hey

We are cleaning up the codebase and improving the project structure for better onboarding. As part of this effort, PR #826 reorganizes src/ from a flat layout into subfolders.

No logic changes — only file moves and import path updates.

What you need to do

Rebase your branch on develop when receiving this comment:

git fetch origin && git rebase origin/develop

Git detects renames automatically. If you get import conflicts, update the paths:

use crate::git;        // now: use crate::cmds::git::git;
use crate::tracking;   // now: use crate::core::tracking;
use crate::config;     // now: use crate::core::config;
use crate::init;       // now: use crate::hooks::init;
use crate::gain;       // now: use crate::analytics::gain;

Need help rebasing? Tag @aeppling

[gghez]

Done, rebased on develop. Had a small conflict in the test section since both branches added tests at the end of diff_cmd.rs — merged both sets, all green.

[KuSh]

Closing this one as overtaken by events rather than as a duplicate — sorry it sat this long.

The headline fix (adding modified to the identical-files guard) has since landed by another route: render_file_diff now branches on diff.changes.is_empty(), which covers modified-only diffs by construction. Verified against the exact repro from #781/#2811 on 0.46.0:

$ printf 'alpha beta\n' > v1; printf 'alpha zzzz\n' > v2
$ rtk diff v1 v2; echo "exit=$?"
exit=1

The branch is also conflicting after the src/ reorganisation you rebased for back in March, so there's nothing here that applies cleanly any more.

The part that did not land is your similarity() rewrite. The current implementation is still Jaccard over character sets, which is order-insensitive — "abcd" and "dcba" score 1.0 — and your prefix/suffix version is strictly better for deciding whether two lines are a modification or an unrelated pair. That's still worth having, and it's small enough to stand alone. A fresh PR with just that change would be welcome.