Skip to content

v0.1.0-beta.6

Pre-release
Pre-release

Choose a tag to compare

@github-actions github-actions released this 02 Jul 09:29

Added

  • Search palette: more results on demand. The palette now shows up to 100 results at once (up from 50), and when there are more, the last row tells you — click show more, or just keep arrowing down past the end, to reveal the next hundred.
  • See highlight colors by name. A new Accessibility setting, Show highlight & shading names in the status bar, reports the stored color names for the text at your cursor (e.g. Hl: Yellow) — so a file's color-coding conventions are readable as text even when the hues are hard to tell apart. Works in the three-pane workspace too.
  • A color-vision friendly palette. New Accessibility setting that remaps the colors that carry meaning — annotation accents, voice-mode dots, the prep timer's Aff/Neg, search-match highlights, category chips — onto a palette engineered to stay distinguishable under red-green and blue-yellow color-vision deficiencies. Works in light and dark themes, and any colors you've set by hand still win.
  • Word footnotes and endnotes are supported. Opening a .docx now keeps its footnotes and endnotes (they previously disappeared on save), shows each as a clickable superscript number — click to read the note, links included, or edit it as plain text — and saving writes them back to Word format. Notes travel with their text when you cut, copy, or send cards. A new Insert Footnote command (unbound by default — assign a key under Settings → Keyboard) creates one at the cursor and opens the note editor immediately; the popover's Delete button (or Backspace over the marker) removes one.
  • Custom dash can trigger on --. The custom dash setting now lets you choose what gets replaced — the classic ---, or just -- for a quicker dash. Backspace right after still reverts to the literal hyphens.
  • Choose what Enter creates after each structural style. Six new Editing settings — one each for Pocket, Hat, Block, Tag, Analytic, and Undertag — control what pressing Enter at the end of that style creates: a normal paragraph (the default, today's behavior) or any structural style. Picking a style acts exactly like pressing Enter and then that style's key, so Tag → Tag starts a fresh card on every Enter.
  • Card sharing recipients can be reordered. Recipient and group rows in Settings → Card Sharing now have up / down arrows, and the Send pill lists them in that order — put the targets you use most on top.
  • Optional undo / redo buttons. A new Appearance setting, Show undo / redo buttons, adds a stacked Undo / Redo pair at the far left of the ribbon. Off by default — the keyboard shortcuts work either way.
  • The timer can sit on either end of the ribbon. A new Appearance setting, Timer position in the ribbon, moves the timer panel to the far right if you'd rather keep the file buttons anchored on the left.
  • The timer is keyboard-drivable. Eight new bindable commands — show/hide the panel, Start / Pause, start each of the three speech presets, start Aff or Neg prep, and Reset — all unbound by default (assign keys under Settings → Keyboard). Apart from show/hide, they're active only while the timer panel is showing.
  • "No color" is now a real pen. Picking No highlight or No background color in the ribbon dropdowns now keeps "none" as your active color (the indicator bar turns white) — the main button, paintbrush mode, and F11 / Mod-F11 then erase wherever you paint, until you pick a color again. Standardize Highlighting / Background with the "no color" pen active removes every highlight or background in scope. Previously the "none" swatch only stripped the current selection.
  • Analytics can be italicized in the nav pane. A new Accessibility toggle, Italicize analytic entries in the nav pane, marks Analytic entries by shape instead of color alone — handy under color-vision deficiencies, and in dark mode or the flat nav list, where the color cue doesn't appear at all.
  • The autosave button shows its real state by shape. When autosave is on but not actually firing (a .docx file, or a doc that's never been saved), the ribbon button is now a hollow chip with a dashed outline instead of a filled one — so "protecting you" vs "armed but inert" no longer differ by color alone.
  • The voice pill always names its mode. The voice status pill now carries a persistent COMMAND / DICTATION / PAINT / ASLEEP badge, so the current mode is readable as text instead of only as the colored dot — previously the mode name flashed briefly and was overwritten by the next thing you said.
  • Annotations distinguishable by shape. A new Accessibility toggle, Distinguish annotations by underline shape, adds a shape-coded underline to each kind of in-document annotation: comments dotted, flashcards solid, AI threads dashed, private notes double — so you can tell them apart even when their tint colors look alike. With the toggle off, annotations are also lighter-weight than before — just the tinted background, no underline — and the tints themselves are slightly stronger to compensate.
  • 20 more colors are now rebindable under Settings → Accessibility → Color overrides: the prep timer's Aff/Neg colors, AI comment / private note / AI repair accents, hyperlinks, the misspelling underline, the due-date dot, find-match highlights, and all ten search category chip colors. Useful if any of the defaults are hard to distinguish.

Changed

  • Settings are easier to navigate. The General, Appearance, and Editing tabs now group related options under section headers (Workspace, Timer, Document typography, Condense, …) instead of one long list, and a new Files tab collects everything about where documents go: new-document defaults, the Send / Read / Marked doc presets, and file search. Several settings also moved to more sensible homes: voice options now sit at the bottom of Accessibility, Timer profile and Timer durations moved to General, and Cite preview on hover and the Flashcards-due dot moved to Appearance.

  • Convert's .zip output is now compressed. When bulk Convert is set to produce a .zip, the archive is now deflate-compressed (it was previously stored uncompressed), so the output is meaningfully smaller.

  • Smaller desktop installer; voice model downloads on first use. The speech recognition model no longer ships inside the installer (which drops its size by roughly 130 MB) — the first time you turn voice control on, CardMirror offers to download it (~130 MB, one time) and tells you when it's ready, so you can keep working meanwhile. If you'll be offline, you can fetch it ahead of time under Settings → Accessibility → Dictation model, which is also where the optional large model lives.

Fixed

  • Performance improvements throughout the editor: the app itself loads about 40% lighter than beta.5 (faster startup on every launch, and a smaller download for the web edition), typing stays smooth with the Find bar open even in very large documents, no more brief stalls after typing pauses in big files, opening Settings no longer leaves the editor slightly slower for the rest of the session, and folder-wide Compress no longer makes the rest of the app unresponsive while it runs.
  • Three-pane: picking a slot with the mouse no longer swallows the next typed 1, 2, or 3. Using the "Open into…" picker with the mouse previously left a hidden key handler behind that ate the next digit typed into the editor.
  • Dark mode: the home-screen file tools (Clean, Convert, Compress) now follow the dark theme instead of rendering as light buttons on the dark background.
  • Hover feedback restored on several buttons that never had a visible one: the per-pane open-file button and doc-stack chip in the three-pane workspace, the slot-picker buttons, the recovery sidebar's Done button, and the card-sharing receive pill.
  • Three-pane: the chip's outline and expand buttons now show their "on" state as a subtle box instead of turning blue. Since outlines are open by default on every pane, the blue glyph read as noise rather than state; the box matches how the focused pane's chip already flags the same thing.
  • First voice use now offers the model download. On a fresh install, turning voice on before downloading the recognition model showed a dead-end error mentioning a developer environment variable instead of offering to download the model. It now offers the download as intended, and the error for genuinely broken installs says to reinstall.