Fit a cluster that draws wider than its cells #29

Merged

natsukium merged 1 commit from fix/fit-clusters into main 2026-10-05 02:04:49 +09:00

Conversation 0 Commits 1 Files changed 6 +719 -128

natsukium commented 2026-10-05 00:52:04 +09:00

Owner

Copy link

Closes #17.

A single glyph wider than its cells is already shrunk to fit them. A multi-codepoint cluster was not: its glyphs were drawn at full size and placed by raw pen advance, so it covered the next character whenever the font could not join what the grid joined.

Changes

• Cluster fit: a cluster whose summed advance overruns its block shrinks so its union ink fits, and is centred. The trigger, slack and ratio are the single-glyph fit's, so a one-glyph cluster fits as its bare char does. All glyphs and the pen shrink by one factor, so marks stay on their base.
  • Covered: 🇦🇦 (no flag glyph), non-RGI ZWJ sequences, 😀🏻, keycaps swash does not ligate, ※́, and a ❤️ left in one cell by a refused VS16 widen.
• Block width: the grid's span (char_span) or OSC 66 w, not the base char's width. Sized-cluster alignment now uses the same span; it undercounted a widened ❤️ and overcounted a refused one.
• Caching: fitted glyphs take their own slot key (SlotKey::FittedGlyph(glyph, px)), because U+0301 can be native in one cluster and shrunk in another. The measurement is memoised in the GlyphIndex, so a font reload drops it; the shape frame carries each cell's fit_px to emission.
• Ligatures: apply_shaped_run now has the overflow check text-shaping.md described. A ligature wider than its source cells (one pixel per cell of slack) is dropped, and the per-character path draws those characters.
• Unfitted clusters at the default Top/Left alignment render exactly as before.

Verification

• just check green.

• Tests on the pinned font set cover fill and centring, the bare-char equivalence, unfitted clusters, the memo, and the pen under an OSC 66 fractional scale. Mock-atlas tests cover the fitted lookup and the grid-span block.

• The tests avoid 🇦🇦: swash panics on it in debug builds (#22).

• Before / after renders (offscreen renderer, Moralerspace Neon HW) of each repro row:

  

Not changed

• ①́ and 字́ still draw their mark past the cell. Their advance does not overrun, so the fit does not apply; the mark's position comes from the font.
• Keycaps and ※́ come out small. swash does not ligate them, so the fit shrinks a whole row of side-by-side glyphs into the cell.

Closes #17. A single glyph wider than its cells is already shrunk to fit them. A multi-codepoint cluster was not: its glyphs were drawn at full size and placed by raw pen advance, so it covered the next character whenever the font could not join what the grid joined. ## Changes - **Cluster fit:** a cluster whose summed advance overruns its block shrinks so its union ink fits, and is centred. The trigger, slack and ratio are the single-glyph fit's, so a one-glyph cluster fits as its bare char does. All glyphs and the pen shrink by one factor, so marks stay on their base. - Covered: `🇦🇦` (no flag glyph), non-RGI ZWJ sequences, `😀🏻`, keycaps swash does not ligate, `※́`, and a `❤️` left in one cell by a refused VS16 widen. - **Block width:** the grid's span (`char_span`) or OSC 66 `w`, not the base char's width. Sized-cluster alignment now uses the same span; it undercounted a widened `❤️` and overcounted a refused one. - **Caching:** fitted glyphs take their own slot key (`SlotKey::FittedGlyph(glyph, px)`), because U+0301 can be native in one cluster and shrunk in another. The measurement is memoised in the `GlyphIndex`, so a font reload drops it; the shape frame carries each cell's `fit_px` to emission. - **Ligatures:** `apply_shaped_run` now has the overflow check `text-shaping.md` described. A ligature wider than its source cells (one pixel per cell of slack) is dropped, and the per-character path draws those characters. - Unfitted clusters at the default Top/Left alignment render exactly as before. ## Verification - `just check` green. - Tests on the pinned font set cover fill and centring, the bare-char equivalence, unfitted clusters, the memo, and the pen under an OSC 66 fractional scale. Mock-atlas tests cover the fitted lookup and the grid-span block. - The tests avoid `🇦🇦`: swash panics on it in debug builds (#22). - Before / after renders (offscreen renderer, Moralerspace Neon HW) of each repro row:  ## Not changed - `①́` and `字́` still draw their mark past the cell. Their advance does not overrun, so the fit does not apply; the mark's position comes from the font. - Keycaps and `※́` come out small. swash does not ligate them, so the fit shrinks a whole row of side-by-side glyphs into the cell.

compare17.png

57 KiB

natsukium added 1 commit 2026-10-05 00:52:04 +09:00

render: fit a cluster that draws wider than its cells ... 1705c03901

A cluster cell rasterized its glyphs at full size and placed them by
raw pen advance, so whenever the face could not join what the grid
joined, the cluster covered its neighbours: a regional-indicator pair
with no flag glyph, a ZWJ sequence the face does not ligate, a skin
tone on a base that takes none, a keycap swash does not ligate, a CJK
`※` with a combining mark. A VS16 widen the grid refused at the last
column left a two-cell emoji in one cell the same way.

Such a cluster now fits by the rule a single overrunning glyph already
follows: its summed advance decides whether it is fitted, the union of
its glyphs' ink decides the size, and it is centred in its block. All
its glyphs and its pen shrink by one factor, so marks stay on their
base. The decision is made where the cluster is shaped, and the frame
carries the fitted size to emission; the fitted glyphs take their own
slot key because a mark such as U+0301 appears native in one cluster
and shrunk in another. Measuring rasterizes every glyph, so the result
is memoised in the glyph index, which a font reload replaces.

The block is the cells the grid gave the cluster, or OSC 66 `w`. The
alignment of sized clusters read the base character's width instead,
which undercounts a widened heart and overcounts a refused one, so it
now uses the same span.

The ligature run gets the overflow check text-shaping.md already
described: a ligature wider than its source cells is dropped and the
per-character path draws them, so a proportional primary face no longer
smears its ligatures over the next cell.

natsukium referenced this pull request 2026-10-05 01:19:41 +09:00

shaping: a VS16 inside a keycap sequence breaks the keycap ligature #30

natsukium force-pushed fix/fit-clusters from 1705c03901 to cd32546a85 2026-10-05 01:47:20 +09:00 Compare

natsukium merged commit edf25b47a0 into main 2026-10-05 02:04:49 +09:00

natsukium deleted branch fix/fit-clusters 2026-10-05 02:04:49 +09:00

natsukium referenced this pull request from a commit 2026-10-05 02:04:50 +09:00

Merge pull request 'Fit a cluster that draws wider than its cells' (#29) from fix/fit-clusters into main

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

Area/ci

CI workflows (.forgejo/workflows/)

  

Area/cli

felis-cli crate, the felis front-door binary

  

Area/client

felis-client crate, the winit GUI frontend

  

Area/client-core

felis-client-core crate, shadow screen, config, connect_or_spawn_daemon

  

Area/daemon

felis-daemon crate, session pool, IPC server

  

Area/deps

Dependency bumps

  

Area/docs

Documentation tree when no narrower area fits

  

Area/grid

felis-grid crate, cells, scrollback, image store

  

Area/manual

User-facing docs: tutorials, how-to, user-facing reference pages

  

Area/nix

Nix flake, home-manager module, package definitions

  

Area/protocol

felis-protocol crate, felis.proto, frame layout, version constants

  

Area/pty

felis-pty crate, PTY backends

  

Area/render

felis-render-wgpu crate, GPU pipeline, atlas

  

Area/shaping

felis-shaping crate, swash shaper, font fallback

  

Area/skills

Agent skills (skills/felis, .agents/skills/)

  

Area/transport

felis-transport crate, framing, socket/pipe, SSH carrier

  

Area/vt

felis-vt crate, VT parser, dispatch

  

Compat/Breaking

Breaking change that won't be backward compatible

  

Kind/Bug

Something is not working

  

Kind/Documentation

Documentation changes

  

Kind/Enhancement

Improve existing functionality

  

Kind/Feature

New functionality

  

Kind/Performance

Throughput, latency, or memory cost

  

Kind/Security

This is security issue

  

Kind/Testing

Issue or pull request related to testing

  

perf-regression-accepted

Passes the instruction-count bench gate for an intended cost; state why in a PR comment

  

Reviewed

Confirmed

Issue has been confirmed

  

Reviewed

Duplicate

This issue or pull request already exists

  

Reviewed

Invalid

Invalid issue

  

Reviewed

Won't Fix

This issue won't be fixed

  

Status

Blocked

Something is blocking this issue or pull request

  

Status

Need More Info

Feedback is required to reproduce issue or to continue work

  

Status

Upstream

Waiting on a change in an upstream project

No labels

Area/ci

Area/cli

Area/client

Area/client-core

Area/daemon

Area/deps

Area/docs

Area/grid

Area/manual

Area/nix

Area/protocol

Area/pty

Area/render

Area/shaping

Area/skills

Area/transport

Area/vt

Compat/Breaking

Kind/Bug

Kind/Documentation

Kind/Enhancement

Kind/Feature

Kind/Performance

Kind/Security

Kind/Testing

perf-regression-accepted

Reviewed

Confirmed

Reviewed

Duplicate

Reviewed

Invalid

Reviewed

Won't Fix

Status

Blocked

Status

Need More Info

Status

Upstream

Milestone

Clear milestone

No items

No milestone

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

felis-terminal/felis!29

No description provided.