max / makeover-layout

State who owns the two corners where the runs meet The renderers disagreed and it looked like drift. makeover-tui splits the shared corners across both tones; makeover-immediate gives them to dark by drawing its dark polyline second. Neither is wrong, and saying so belongs here rather than in either renderer, since a cross-renderer rule is what this crate is for. The corners belong to both edges. What a renderer can do about that is bounded by resolution: a terminal cell is roughly 8x17 device pixels, so handing the whole corner to one tone thickens that edge by a visible cell, and a half-cell glyph has room to divide it. A one-point pixel stroke makes the corner a one-point square, where a diagonal seam is sub-pixel and antialiasing resolves it to the blend a mitred join already produces. Same rule, one resolution where it has something to say and one where it does not.

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

Author: Max Johnson <me@maxj.phd> · 2026-07-30 17:27 UTC

Signed with PGP, not checked

Commit: 4d575665c75366b3cfee25bd78f83486d4d7a992

Parent: 1e805dd

1 file changed, +22 insertions, -0 deletions

M src/lib.rs +22

			@@ -86,6 +86,28 @@
86	86		/// light is always assumed to come from the top left: every consumer measured
87	87		/// agreed on that and none of them ever varied it, so it is an invariant here
88	88		/// rather than a parameter.
	89	+	///
	90	+	/// # The two corners that belong to both edges
	91	+	///
	92	+	/// Top-right and bottom-left are where the lit run meets the shaded one, and
	93	+	/// the description's claim is that they belong to both. How a renderer says
	94	+	/// that is its own business, because the answer is bounded by resolution and
	95	+	/// not by taste:
	96	+	///
	97	+	/// - A terminal cell is roughly 8x17 device pixels, so giving the whole corner
	98	+	/// to one tone thickens that edge by a cell and reads as one run overrunning
	99	+	/// the other. A half-cell glyph divides the cell already, so `makeover-tui`
	100	+	/// splits it and recovers real information. Its box-drawing fallback cannot:
	101	+	/// a single stroke has no half to give, so there both corners go to dark.
	102	+	/// - A pixel bevel is a one-point stroke by default, which makes the corner a
	103	+	/// one-point square. There is nothing to divide — a diagonal seam across one
	104	+	/// point is sub-pixel, and antialiasing renders it as the blend a mitred join
	105	+	/// already produces. So `makeover-immediate` mitres and is not diverging;
	106	+	/// it is the same rule at a resolution where the split degenerates.
	107	+	///
	108	+	/// Stated here so the difference reads as a decision rather than as drift. A
	109	+	/// renderer with room to divide the corner should; one without should mitre or
	110	+	/// pick the shaded tone, and neither is a bug.
89	111		#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
90	112		pub enum Bevel {
91	113		/// Lit from the top left: light on top and left, dark on bottom and right.