umputun / revdiff / 25546184661

90%

push

github

feat: decouple mouse wheel from diff cursor (#173)

* feat: decouple mouse wheel from diff cursor

wheel events on the diff pane now scroll the viewport only.
the diff cursor stays on its current logical line and is
pinned to the visible edge if scrolling pushes it off-screen.
matches less/vim mouse semantics.

before, every wheel notch advanced the cursor by the wheel
step and dragged the viewport along to keep the cursor row
stable. under trackpad momentum that felt automatic and
slightly inertial because each event re-rendered the diff to
move the cursor highlight. the new path skips the re-render
when the cursor stays in place, so plain scrolling is just a
viewport offset change.

* style: gofmt cleanup in test files

trim trailing newline in config_test.go and re-align struct
field paddings in compare_test.go to match gofmt output.
no behavior change.

* fix(review-loop): iteration 1 addressed 5 findings

- [major] app/ui/mouse.go:pinDiffCursorTo: fix cursor-invisible bug in wrap mode —
  overlap check now uses cursorTop (marker row) instead of cursorBottom; when the
  cursor line straddles the viewport top boundary, advance past the continuation
  rows using cursorBottom+1 so the next line marker is actually visible
- [major] site/docs.html:560: sync scroll-wheel description to match README and
  plugin reference docs (viewport-only behavior with cursor pinning)
- [minor] app/ui/mouse.go:scrollDiffViewportBy: fix godoc - "clamp boundary"
  replaced with accurate "clamped target equals current offset"
- [minor] app/ui/mouse_test.go: add WheelUp_PinsCursorToBottom (bottom-pin branch),
  WheelDown_NoopAtMaxOffset (maxOffset no-op), WheelInWrapMode_PinsToFirstVisibleLine
  (wrap-mode regression guard)

* fix(review-loop): final cleanup pass — addressed 1 test-gap finding

- [test-gap] app/ui/mouse_test.go: replace weak WheelInWrapMode_PinsToFirstVisibleLine
  test (did not exercise straddle branch) with two focused tests:
  WheelInWrapMode_CursorAbo... (continued)

36 of 42 new or added lines in 1 file covered. (85.71%)

9688 of 10800 relevant lines covered (89.7%)

3198.88 hits per line