max / balanced_breakfast

Ask what is pointing at the screen, not what hardware exists The byte-identical twin of goingson's touch detection, fixed in the same pass so the duplication does not outlive the bug. The sniff asked the hardware inventory, so a laptop with a touchscreen and a mouse came back touch; it now asks matchMedia with makeover-geometry's Density::Touch condition, the query the generated geometry.css already uses. No build-side guard here, unlike goingson: this build.rs has no check pass at all, and copying one in is the wrong shape when the checker belongs in makeover-build. Tracked there.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: Max Johnson <me@maxj.phd> · 2026-08-10 22:15 UTC

Signed with PGP, not checked

Commit: 04321714b43ba9ed30068b44f4cc42301e004761

Parent: df5e04c

1 file changed, +10 insertions, -1 deletion

M src-tauri/frontend/js/touch.js +10 -1

			@@ -9,7 +9,16 @@
9	9
10	10		// Touch Detection
11	11
12		-	const isTouchDevice = ('ontouchstart' in window) \|\| (navigator.maxTouchPoints > 0);
	12	+	// The density question, asked the way geometry.css asks it. This was a
	13	+	// capability sniff off the window and the navigator until 2026-08-10: it
	14	+	// asked what hardware exists rather than what is pointing at the screen,
	15	+	// so a laptop with a touchscreen and a mouse came back touch and lost its
	16	+	// hover affordances. The string is makeover_geometry::Density::Touch's own
	17	+	// media condition. Unlike goingson, nothing here fails the build if it
	18	+	// drifts -- BB's build.rs has no check pass yet.
	19	+	const TOUCH_DENSITY = '(hover: none), (pointer: coarse)';
	20	+
	21	+	const isTouchDevice = window.matchMedia(TOUCH_DENSITY).matches;
13	22
14	23		// Long Press
15	24