Conversation
…rnate screen The panel's wheel handler always scrolled the emulator's own history, so a full-screen TUI got nothing: Claude Code runs in the alternate screen with SGR mouse tracking on, and the wheel was dead inside it (zeronsh#361). A native terminal reports the wheel to such a program instead. `Emulator::wheel_route` reads the modes the program set and picks one of three destinations, in xterm's precedence: mouse reporting (DECSET 1000/1002/1003, encoded per 1006/1005/X10), alternate screen with alternate-scroll (1007, on by default) as cursor keys honoring DECCKM, or the scrollback as before. The panel encodes with `wheel_report_bytes` / `wheel_arrow_bytes` and writes through the existing input coalescer. Shift+wheel always reaches the scrollback, and a mouse-mode program on the primary screen gets no report while the view is scrolled into history — the wheel walks back to the live bottom first, where the reported coordinates mean something. Wheel lines now pass through a `WheelAccumulator` on every route: a trackpad's sub-line pixel deltas used to round to nothing, and a partial line must not leak from the scrollback into a stray key press when the route changes.
…scrollback Review follow-ups on the wheel forwarding: - `wheel_lines` reads a Shift+wheel back from the X axis when Y is empty: AppKit moves a physical mouse's Shift+wheel there (that is how horizontal scrolling works for mice) and gpui passes the raw deltas through, so the "Shift always reaches the scrollback" rule was a silent no-op on macOS. - A tab whose program has exited routes to the scrollback regardless of the modes it left behind; `queue_input` already dropped the bytes, which left the wheel doing nothing. - `effective_wheel_route` gathers the three overrides (Shift, exited, scrolled into history under mouse mode) into one pure function with its own tests, instead of a closure inside the gpui listener. - `GridGeometry::cell_at` replaces the six-argument `cell_at` call that `grid_point_at` and the wheel handler both spelled out, and gets a test that follows a pointer through the placement into an SGR/X10 report.
5 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Emulator::wheel_route,wheel_report_bytes,wheel_arrow_bytes, and aWheelAccumulatorso trackpad sub-line pixel deltas add up to whole lines instead of rounding to nothingCloses #361.
Root cause
on_scroll_wheelalways calledTerm::scroll_display. A full-screen TUI in the alternate screen has no history to scroll, and one with mouse tracking on expects the wheel as input — Claude Code does both (?1049h,?1000h,?1006h), so the wheel was dead inside it while a native terminal scrolls fine.Out of scope
MOUSE_REPORT_CLICK,MOUSE_DRAG,MOUSE_MOTION): the same encoder applies, but clicks currently drive text selection and need their own design.Testing
cargo test -p zeron-ui terminal::(67 passed, 9 new: mode routing over real escape sequences, the Shift/exited/history overrides, pointer → placement → SGR/X10 report, Shift axis fallback, SGR/X10/UTF-8 encoding incl. the 223/2015 caps, arrow bytes, accumulator)cargo test -p zeron-ui(967 passed)rustfmt --edition 2024 --check crates/ui/src/terminal/view.rs crates/ui/src/terminal/emulator.rs;panel.rscarries one pre-existingon_terminal_hoverhunk frommainthat this change leaves untouchedcargo clippy -p zeron-ui --all-targets: no findings undercrates/ui/src/terminal/scripts/run-macos-dev.sh: wheel and trackpad scroll the conversation insideclaudein the right-sidebar terminal; a plain shell scrolls its scrollback as before