Repository navigation
Directives: glyphs and autocomplete
The last two phases of the directive seam. Both are inert until you register a directive.
- Glyphs. A self-contained call (
@pagebreak,@glyph(star.fill)) can draw as an SF Symbol, replacement text or an image, supplied by the directive'spresentation. The source stays in storage and collapses to zero width like inline LaTeX, so selection, find, copy and undo still see it, and the caret reveals it. - Autocomplete for directive names (
@fo) and argument values (@icon(sta), viaonDirectiveCompletionandpendingDirectiveCompletion. The engine detects the trigger, ranks the candidates and routes ↑/↓/↵/Esc; the embedder draws the list. Closed keyword sets and booleans complete from the schema alone;valueCompletions(for:prefix:)covers dynamic domains.
Demo/ adds @icon, @flag, @emoji and @pagebreak as embedder-side examples.
Added
SpellCheckingPolicy.automaticQuoteSubstitution(defaulttrue) keeps straight'and"for embedders editing raw Markdown or LaTeX.
Fixed
- Nested lists keep their levels when copied out and when pasted in from Mail or Notes; task boxes render with the list helpers off.
- A trackpad held against the top or bottom edge no longer flickers.
- After a programmatic content swap, the code-block cache no longer reads the previous document's ranges, which could put a copy button over plain prose.
Changed
- Raw source mode reports no code blocks.
Full changelog: 0.13.0...0.14.0
Contributors
by @wildthink — directive glyphs (#158) and autocomplete (#159).
by @xVanTuring — configurable smart quotes (#189).
by @secondstagehq — the stale code-block cache after a content swap (#162).
by @luca-chen198 — nested list levels and task boxes (#183, #184), and the trackpad edge flicker (#194).