Merge conflict corpus and benchmark

[flyingrobots]

Migrated from Method backlog

GitHub Issues are now the live work tracker. Repository docs remain Method evidence.

Field	Value
Source backlog	docs/method/backlog/up-next/DX_merge-conflict-corpus.md
Archived source	docs/archive/backlog/github-issue-migration-2026-06-01/docs/method/backlog/up-next/DX_merge-conflict-corpus.md
Original lane	up-next
Original id	DX_merge-conflict-corpus
Original legend	DX
Original feature	merge-strands-worldlines

Original backlog card

Merge conflict corpus and benchmark

The merge-geometry work needs a grounded benchmark corpus, not just nice
examples. We should gather a set of real conflicts from this repo and nearby
repos and classify them.

Questions to answer:

• How many conflicts are really projection conflicts?
• How many are genuine semantic conflicts?
• How many are governance conflicts?
• Which conflict classes are worth attacking first?

Work:

• collect 50-100 real merge conflicts or hand-resolution cases
• normalize them into a reproducible fixture corpus
• label each case by conflict class and domain
• measure how many disappear under simple structured lifting
• use the corpus for CLI, debugger, and future merge-engine tests

Source

• merge-geometry discussion, 2026-04-09
• docs/design/merge-lifting-worked-examples.tex