git bisect start <bad> <good>     # or: start; bad; good <tag>
# git checks out midpoint; you test, then:
git bisect good | bad | skip
git bisect reset                  # finish, back to original HEAD

git bisect start HEAD v1.4.0
git bisect run ./reproduce.sh
git bisect run sh -c 'cargo build && cargo test --test regression'

git bisect log > bisect.log
git bisect reset
# later:
git bisect replay bisect.log

git blame -L 40,80 file.py                 # line range
git blame -L :funcName:file.py             # by function
git blame -w file.py                       # ignore whitespace-only
git blame -C file.py                       # detect moves within same-commit diffs
git blame -C -C -C file.py                 # across files in any commit (thorough)
git blame --ignore-rev <sha> file.py       # skip a mass-format commit
git config blame.ignoreRevsFile .git-blame-ignore-revs

git log -S"token"                  # when token count in file changed
git log -S"pat" --pickaxe-regex
git log -G"regex"                  # any line matching was touched
git log -L :ClassName:src/p.ts     # full evolution of a symbol
git log --follow -p -- path        # history across renames

git log -S"oldLogic" -- src/       # last clean commit
git bisect start HEAD <that-sha>
git bisect run pytest tests/test_regression.py