Skip to content

Releases

0.16.1

Choose a tag to compare

@github-actions github-actions released this 15 Nov 04:41
f9ffe62

Fixed

  • With Vim Motions' Vim in text areas on, Ctrl/Cmd+Enter saves the card editor again, and a save made right after typing keeps the last letters typed. Vim Motions puts its own editor over the editor's text boxes, keeps Ctrl/Cmd+Enter to itself, and copies its text back only after a short pause. (#10)

0.16.0

Choose a tag to compare

@github-actions github-actions released this 15 Nov 04:22
52dd8f4

Added

  • Save the card being edited is a command you can give a hotkey of your own in Obsidian's Settings → Hotkeys. That hotkey saves the card editor alongside Ctrl/Cmd+Enter, for when another plugin, such as a Vim plugin, takes Ctrl/Cmd+Enter. (#10)

0.15.0

Choose a tag to compare

@github-actions github-actions released this 13 Nov 08:37
a2af4a7

Added

  • On a computer, E opens the editor for the card you are reviewing, on the question or the answer side. Ctrl/Cmd+Enter saves and Escape discards, as before. With Custom hotkeys on, Edit this card is a command you can give a hotkey of your own. (#10)

0.14.2

Choose a tag to compare

@github-actions github-actions released this 13 Nov 04:49
3b13ebb

Changed

  • In the Studio look, a cloze card reads the same before and after its answer shows: the same size and place, left-aligned (right-aligned for right-to-left text), without the Answer label, so only the hidden words change. Ordinary cards keep their large centred question and the label. (#11)
  • The Studio screens take their page, card and inset backgrounds from your Obsidian theme, in light and dark, and the glow behind the review screen is gone. The accent and the rating colours stay the Studio's own. (#11)

0.14.1

Choose a tag to compare

@github-actions github-actions released this 07 Nov 18:13
5adf2ee

Fixed

  • With AI checks typed answers, an exam sends its typed answers only to the AI provider it started with. If another provider is set up before you submit, nothing is sent and every typed answer keeps its exact mark, with the reason in the results. An OpenAI-compatible server at another address is another provider, and so is http:// instead of https:// on the same host. The checks of one submit all go to the provider they started with, even if the settings change while they run. An exam started in 0.14.0 with AI checks did not keep its provider, so if you submit it after updating, its typed answers are not sent.
  • Retake, the last exam's setup and Take an exam from this note turn AI checks typed answers on only when the AI provider of the last exam is still the one set up.
  • A retake of an exam from notes stays on Notes when some or all of its notes were renamed, deleted or have nothing left to ask. The notes that are gone are named, and with none left Start exam waits for you to add a note instead of taking every deck.
  • The exam setup lists your unfinished exams even when there are no cards left to make a new exam from.
  • When AI checks stop (Show results now, the time limit, or failures in a row), an answer sent with no verdict back says so, and one never sent says it was not sent, instead of both saying the checks stopped before it.

0.14.0

Choose a tag to compare

@github-actions github-actions released this 05 Nov 01:58
bf8cb3a

Added

  • Take an exam from chosen notes. The exam setup can take its cards from Notes instead of decks, with Add note to search the notes that have cards to ask. Take an exam from this note, in the command palette and in a note's file menu, opens the setup with that note chosen. Retake, the last exam's setup and a resumed exam keep the notes. (#6)
  • AI checks typed answers in exams. When AI is set up for card generation, the exam setup offers to check typed answers that do not match exactly with the same meaning check as Check meaning with AI, when you submit. One with the same meaning counts as right; the results and the exam's file show the AI's verdict and reason for each answer it checked. Only the question, the expected answer and your typed answer are sent, and nothing is sent unless you turn it on. (#6)

Fixed

Studying and scheduling

  • Daily limits count from your start of day, not from midnight. With a 04:00 start, cards studied at 01:00 no longer use up the next day's limit. The reminders and custom study's "forgotten in the last days" count the same way.
  • With SM-2, a new card answered Again no longer uses a review from the daily review limit, and is no longer dropped from the session when the review limit is reached. With the "Any card from any deck" order it now comes back before the session ends.
  • Undo with "Bury sibling cards" on puts the siblings back in the session.
  • FSRS load balancing keeps Hard before Good and Good before Easy; they could get the same day.
  • The FSRS optimizer leaves out cards whose history does not start with their first answer, such as cards carried over from the Spaced Repetition plugin. They made the first intervals of new cards far too long.
  • Settings that sync in from another device while Obsidian starts are applied. A synced settings change no longer makes new SM-2 cards ignore the ease of their note.

Review screen

  • Closing a review while it reloads (after Review mistakes, or an undo that reloads) no longer leaves a hidden session that takes the keys of later reviews.
  • Starting a session from the waiting screen no longer leaves the new card unanswerable.
  • After editing a multiple choice card on its answer side, the answer key and explanation shown are the edited ones.
  • Undoing a reset no longer removes another answer from the session's progress and counts.
  • The status bar counts a card that is in two decks once.

Editing cards and notes

  • The edit dialog keeps the indentation of the answer, so code blocks and nested lists survive a save, and Save without a change writes nothing.
  • The edit dialog no longer saves line breaks that the note would read as another card. It says why instead.
  • Saving an image occlusion after an empty card-metadata callout no longer deletes or changes text further down the note.
  • With folders as decks, answering a card keeps the tag in front of its question.
  • Deleting scheduling data with its tags removes only those tags: deleting #review no longer also removes #reviewed.

Settings

  • A cloze pattern the plugin cannot read, such as {{answer}}, is refused with a message saying what it needs. One saved before is dropped when the settings load, so the vault scan no longer fails on every start.
  • Importing settings from Spaced Repetition keeps its hidden file-menu items, its random card order and its old cloze switches.

Statistics and exams

  • With a start of day other than midnight, the forecast no longer shows SM-2 cards a day early.
  • Closing Statistics while it loads no longer leaves its charts running.
  • Closing the FSRS optimizer while it loads no longer freezes Obsidian for the whole optimization.
  • The exam setup no longer slows down with every click when there are many decks.
  • Last exam on the home shows the newest exam when two end in the same minute, and "today" and "yesterday" in the exam lists follow the day, not 24 hours.
  • A double click on Apply in custom study raises the limit once.

Import and AI

  • Importing a package with thousands of media files is much faster.
  • With the single-line style, a basic note whose front ends with : is no longer turned into a reversed pair.
  • Anki decks whose names differ only in symbols, such as C, C++ and C#, stay separate decks.
  • A note with a cloze number above 500 is skipped instead of freezing the import.
  • Long lists of atomic clozes are read quickly again: 2,000 lines took 17 seconds.
  • With an end marker and no start marker, generated multi-line and cloze cards are written with the end marker, so they no longer take in the next section of the note.

Security

0.13.0

Choose a tag to compare

@github-actions github-actions released this 28 Oct 19:46

Added

  • Check meaning with AI for typed answers (#4). Turn it on in Settings → Studying, under Type your answers. When a typed answer on a basic or reversed card does not match exactly, the AI provider you set up for card generation checks whether it means the same, and says "AI: same meaning", "AI: different meaning" or "AI could not decide" under the answer. On "same meaning" the suggested button moves to Good. You still choose the rating; the AI never rates for you.
    • Off by default. Nothing is sent unless you turn it on; then only the question, the expected answer and your typed answer are sent.
    • A changed number, a flipped "more than"/"less than", or a "not" on only one side can never count as the same meaning.
    • The answer and the buttons show at once; the check runs beside them and gives up after 10 seconds. Cloze, image occlusion, multiple choice and exams keep the exact check.

0.12.2

Choose a tag to compare

@github-actions github-actions released this 21 Oct 04:13

Fixed

  • Hovering over a card during review no longer shows a "Flashcard" tooltip that covered the answer buttons, in both the Classic and Studio looks. Screen readers still announce the card as "Flashcard". (#2)

0.12.1

Choose a tag to compare

@github-actions github-actions released this 13 Oct 05:06

Fixed

  • Updated the bundled brace-expansion library (through minimatch) to a version without the published advisory. Build and test tools were updated too.
  • The stylesheet follows Obsidian's plugin review rules: no !important and no :has() selectors, so themes and CSS snippets can override Flashcard Studio's styles more easily. Nothing looks different.

0.12.0

Choose a tag to compare

@github-actions github-actions released this 12 Oct 21:46

Changed

  • Flashcard Studio has its own command ids and view names. Custom hotkeys for Flashcard Studio commands need to be bound again once.
  • Open Flashcard Studio tabs are reopened after a restart.
  • The CSS classes of the plugin start with fs- now, not sr-. CSS snippets and themes that target the old sr-* classes must use the fs-* ones.

Fixed

  • Undo restores the note in one step, so an edit made at the same moment is not lost.
  • Reading a note no longer rewrites it to delete schedule segments it has too many of; they stay in the note.
  • Settings changed on disk by a sync service are applied without a restart.
  • Removing a card from a question with several cards says that the whole question goes, and a note that changed in the meantime is reported instead of silently losing the card.
  • Asking for the decks while a scan runs gives the state after the request, so a note written during the scan is not missed.
  • Delete scheduling data only writes notes it changes, and leaves tags in code blocks and inside words alone.
  • Moving schedules into callouts leaves code blocks, tables and lists intact and says how many it left.
  • Postpone moves only the cards it counted.
  • A warning when an OpenAI-compatible server is reached over plain http://.