FA-75613 / Text diff and three-way merge / Member archive

Rename similarity pairing: candidates are paired in path order instead of by score · case 03

A weaker pairing claims a file before its best match is considered.

Member previewVariant 3 · 3 implementations · 12 checks per implementation

Case contract

For deleted and added files (path -> lines), a pair scores floor(100 * multiset-common-lines / max(len_a, len_b)); empty files never pair. Pairs scoring at least the threshold are taken greedily by descending score, ties by (deleted path, added path) ascending, each file used at most once. Return [deleted, added, score] triples sorted by deleted path.

Why this case matters

Rename and copy detection decides whether a delete plus an add is shown as a rename with a small diff.

One recorded failure

Sample boundary fixture

This sample comes from the broken implementation of a controlled reproducer.

Boundary fixture	Actual	Expected	Outcome
lower score pairs after higher ones	`[["x", "p", 90], ["y", "q", 50]]`	`[["x", "q", 100], ["y", "p", 50]]`	Failed

MEMBER ARCHIVE

The complete case is available to members.

This record includes three runnable implementations, regression fixtures, execution results, and source hashes.

Member access is invitation-based. Sign in with your invited account to inspect the sources.