Plex Math: pieces of horizontal stretchy brackets and double arrows snap to different pixel rows
Personne n'a encore pris cette issue.
Évaluation
- Difficulté
- 4/5
- Temps estimé
- 3-5 jours
- Accessibilité débutants
- 46/100
- Type d'issue
- Bug
- Clarté
- Plutôt claire
- Activité
- Active
- Stack technique
- tex
- Domaine
- design
Piste de recherche
The issue identifies uni23DE.lft, uni23DE.ext and uni23DE.mid and reports similar joins in other horizontal assemblies; it does not name source files or tests. Start by locating the font source and instructions for these glyphs, then reproduce the reported joins at the listed sizes with hinting enabled. Done means the shared strokes align across the tested assemblies without losing crispness or creating brighter pixels at joins.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Description
In IBM Plex Math 1.1.0 (@ibm/plex-math on npm), the pieces of several horizontal stretchy glyphs hint the same stroke with different programs. At many sizes the hinted stroke of the extender lands exactly one pixel above or below the stroke of the end or middle piece next to it, so the assembled glyph has a step at each join. I see this in Chrome and Firefox on Windows on a 1080p screen, both at text sizes and under pinch zoom.
U+23DE (top curly bracket) is an example. The outlines of uni23DE.lft, uni23DE.ext and uni23DE.mid all put the stroke at y = 714 to 780. lft and mid have their own instructions (44 and 68 bytes). ext is a reference to uni23B4.ext, the extender of U+23B4, and uses its 11 bytes of instructions. The three programs round the shared stroke to different pixel rows at some sizes.
Here is U+23DE at 20 px, rendered with FreeType with hinting on and magnified. The top row is the original font; the bottom row is the same font rehinted with ttfautohint, discussed below.
I rendered every horizontal assembly in the MATH table with FreeType, hinting on, from 14 to 80 px in 0.25 px steps, and compared the stroke height on both sides of each join. These have 1 px steps:
| Glyphs | Sizes with a step (of 265) |
|---|---|
| U+21DA, U+21DB (triple arrows) | 83 |
| U+23B5, U+23DD, U+23DF, U+23E1 (bottom brackets) | 63 |
| U+23B4, U+23DC, U+23DE, U+23E0 (top brackets) | 52 |
| U+21D0, U+21D2, U+21D4 (double arrows) | 24 |
The single arrows, harpoons and combining marks have none. I didn't test the vertical assemblies: Windows and FreeType apply hints only in y, so their vertical strokes shouldn't be affected.
To reproduce, open this page on Windows at 100% scaling and look along the rows for steps in the brace:
<!doctype html>
<meta charset="utf-8">
<style>
@font-face { font-family: "IBM Plex Math"; src: url("https://cdn.jsdelivr.net/npm/@ibm/[email protected]/fonts/complete/woff2/IBMPlexMath-Regular.woff2") format("woff2"); }
math { font-family: "IBM Plex Math", math; }
</style>
<div id="rows"></div>
<script>
const sample = `<math><mover><mrow><mi>a</mi><mo>+</mo><mi>b</mi><mo>+</mo><mi>c</mi><mo>+</mo><mi>d</mi></mrow><mo stretchy="true">⏞</mo></mover></math>`;
for (let px = 14; px <= 22; px += 0.25)
document.getElementById("rows").insertAdjacentHTML("beforeend", `<div style="font-size:${px}px">${px}px ${sample}</div>`);
</script>
What I tried on a local copy:
- Removing all instructions makes the pieces line up, but the strokes get blurry at text sizes and the overlapping connectors show as brighter pixels at the joins.
- Rehinting the whole font with ttfautohint at default settings removes the steps at all 265 sizes, but the stroke keeps its real thickness (about 1.3 px at 20 px) and turns grey across rows, as in the bottom row of the image.
- With ttfautohint's strong stem snapping the stroke is crisp more often, but the steps come back (72 sizes for U+23DE).
So I don't think a generic autohinter can fix it. The pieces of each assembly would need to snap the shared stroke edges the same way, for example by rounding them against the same CVT entries.
- Langage dominant
- CSS
- Étoiles
- 11.8k
- Forks
- 626
- Merge moyen
- 1 j 9 h
- PR mergées (30 j)
- 2
Préparer son environnement
Ce projet ne fournit ni conteneur de développement, ni Dockerfile, ni guide de contribution : l'installation est à votre charge. Commencez par son README, et consultez notre guide de la première contribution pour les étapes générales.
Par où commencer
- Lisez l'issue en entier, puis le guide de contribution du projet.
- Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
- Forkez le dépôt et travaillez sur une branche.
- Ouvrez une pull request qui référence le numéro de l'issue.
Autres issues de IBM/plex
-
Adding Hanja in Plex KROuverte
Difficulté 5/5 Plus d'une semaine Accessibilité débutants 15/100
-
security
Difficulté 2/5 1-3 heures Accessibilité débutants 35/100
-
Difficulté 4/5 1-3 heures Accessibilité débutants 8/100
-
Difficulté 4/5 3-5 jours Accessibilité débutants 40/100
-
iBM Plex Sans ArabicOuverte
Difficulté 3/5 1-2 jours Accessibilité débutants 55/100
Issues similaires
-
accessibility
Difficulté 1/5 Moins d'une heure Accessibilité débutants 78/100
numerique-gouv/lasuite-landingpage#304 ·
Les mainteneurs répondent en général sous 1 jour
-
ExtrudeOuvertearea:kernel bug needs-triage
Difficulté 2/5 1-3 heures Accessibilité débutants 62/100
bherbruck/solvecraft#16273 · 1 commentaire · 1 réaction ·
-
.Needs Triage Priority:P3 Type:Bug
Difficulté 2/5 1-3 heures Accessibilité débutants 76/100
metabase/metabase#87117 · 1 commentaire ·
Les mainteneurs répondent en général sous 1 jour
-
enhancement
Difficulté 2/5 1-3 heures Accessibilité débutants 75/100
catppuccin/tmux#608 ·
-
Difficulté 2/5 1-3 heures Accessibilité débutants 72/100
ocftw/open-star-ter-village#507 · 1 commentaire ·
Les mainteneurs répondent en général sous 1 jour