Content is chosen from a catalogue rather than handed over - #48
Open
vgomx wants to merge 2 commits into
Open
Conversation
Every install wrote the whole library: twenty-two decks and a hundred and sixty-five cards across four languages, plus all twenty-four grammar notes, most of it in languages the reader had not picked and would never open. That is the opposite of a catalogue. There is nothing to add when everything is already there, and no way to tell what you chose from what you were given. A fresh install now arrives with one starter pack in the language it opens in — measured on a wiped database: one deck, six cards, one note, one language. A pack is a rule, the words that exercise it, and the verbs to drill it on. That shape was already in the app and unconnected: the notes are the grammar themes, six per language and exactly the ones people trip over — de or het, ser or estar, what triggers the subjunctive, separable verbs come apart — and the decks are the practice. A pack pairs them. It is a view over that content rather than a second copy. A pack names a deck id and note ids, and installs them through the same buildDeck the first seed uses, so a pack added months later writes the records a fresh install would and cannot drift from the material it offers. The verbs are a focus list, not data. The drill already loads every verb the language has — NL, ES and PT ship whole tables — so what a pack contributes is which of them the theme is about. It narrows the pool rather than reweighting it, because the point of the pack is that it is these verbs, and the distractors still come from the whole language so a wrong answer does not give itself away. Checked: with Separable verbs installed, the drill asks about ophouden. Four of the eight Dutch lemmas I first wrote — opbellen, uitgaan, meedoen, opruimen — are not in the table Wiktionary gave us. They are replaced with verbs that are. A focus list naming verbs the drill cannot ask about is a pack that quietly does nothing. Nobody loses anything. ensureSeeded is still guarded on the database being empty, so an existing library is untouched, and a pack whose deck is already on the shelf reads as added — a weaker claim than the packs store makes, and the right one for a database seeded before the catalogue existed. Verified by clearing the store against two decks: two marked added, three still offered. ensureNotesSeeded is now a no-op rather than deleted, so a database seeded by an older build keeps every note it has. Notes arrive with the pack that explains them. The store goes to v6 and the backup to format 6, both carrying the packs; restoring decks without them would offer back what the reader had already picked. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three things the catalogue was wrong about, all found by looking at it. **Decks, not packs.** The reader is offered a deck: that is the word this app already uses for a set of cards you practise, and a second word for the thing that becomes one would be a word to learn for something they can already name. The code keeps `Pack` for the offer, because `Deck` is the record it installs and one of them has to give. **They were thin.** Six to eight cards is a deck you finish before it teaches you anything. 165 cards across 22 decks becomes 301, none of them under ten, most between fourteen and seventeen. Two things the checks caught rather than I did. Forty-one of the cards I wrote were words those decks already had — whereas, sin embargo, de collega, ficar — and the later copy is dropped, so the originals and their illustrations stand. And no card names an illustration that does not exist, because the palette was built by reading public/openmoji first rather than by guessing and finding out. **New deck was desktop-only.** The plus sat in the deck sidebar's header, and a phone never shows that sidebar — so on an app whose point is the cards you write, writing one was possible at a desk and nowhere else. That is the honest answer to "why can't the user add their own": on a phone they could not. It is in the top bar now, which every screen and every size has. 44px under a thumb and 36 under a pointer, rather than the 14px glyph it was. Only inside Flashcards and Review, though — a button that makes a deck while you are reading an etymology has wandered away from the thing it acts on. Measured across five routes and two widths: present on the four where it belongs, absent on the two where it does not. The deck list's empty state names both ways in as well, since the catalogue is a starting point rather than the only source of cards. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
How content was offered
Every install wrote the whole library: 22 decks, 165 cards across four languages, and all 24 grammar notes — most of it in languages the reader had not picked and would never open. Your Dutch decks were not yours because you chose Dutch; they were the rows every install starts with, alongside Portuguese, English and Spanish.
That is the opposite of a catalogue. There is nothing to add when everything is already there, and no way to tell what you chose from what you were given.
What a pack is
A rule, the words that exercise it, and the verbs to drill it on.
That shape was already in the app, unconnected. The grammar notes are the theme list — six per language, exactly the ones people trip over: de or het?, ser or estar?, What triggers the subjunctive?, Separable verbs come apart. The decks are the practice. A pack pairs them.
It is a view over that content, not a second copy. A pack names a deck id and note ids and installs them through the same
buildDeckthe first seed uses, so a pack added months later writes the records a fresh install would, and cannot drift from the material it offers.22 packs, every id resolving, every seeded deck claimed exactly once.
The verbs are a focus list, not data
The drill already loads every verb the language has — NL, ES and PT ship whole conjugation tables. So a pack does not carry forms; it says which verbs the theme is about. It narrows the pool rather than reweighting it, because the point of the pack is that it is these verbs — and the distractors still come from the whole language, so a wrong option does not give itself away by its stem.
Four of the eight Dutch lemmas I first wrote from memory —
opbellen,uitgaan,meedoen,opruimen— are not in the table Wiktionary gave us. They are replaced with verbs that are. A focus list naming verbs the drill cannot ask about is a pack that quietly does nothing.A fresh install
One starter pack, in the language it opens in. Measured on a wiped database: 1 deck, 6 cards, 1 note, 1 language.
Nobody loses anything
ensureSeededis still guarded on the database being empty, so an existing library is untouched.ensureNotesSeededis a no-op rather than deleted, so a database seeded by an older build keeps every note it has. Notes now arrive with the pack that explains them.Checked
ophouden— one of the pack's eight, out of thousands in the tableIndexedDB goes to v6 and the backup to format 6, both carrying the packs: restoring decks without them would offer back what the reader had already picked.
Known
The catalogue adds but does not remove. Deleting the deck already exists and is the honest way out; a Remove that took back cards you have been grading is a different and more dangerous feature.
Every pack's cards come from the existing seed. The themes with a note but no matching deck — Dutch
er, the Spanish personala— ride along with a related pack rather than having practice of their own. New card sets are the next content job, not this one.