halkyon/apps/learn/halkyon_learn_server
Repository files (latest commit first)
Filename Latest commit message Latest commit date
Tizian 「ティツィアーン」 358d983011
refactor(learn): own rows name shared content by its ID alone
ADR 0044, D26. The sync engine links no row of one space to a row of
another: a reference across spaces is blanked, on the device and on the
server alike (found by experiment, 11 Nov 2026). Reviews, card, note and
path tags, PDF readings and path answers are the person's own (D2, D25)
but will name cards, notes, paths, steps and PDFs of spaces they share,
so those six tables keep their columns without the relation. Every
gateway already deletes them as it deletes what they name.

The migration drops the nine foreign keys of those tables and adds the
three to `tag` again, which Serverpod numbers anew; no column or row
changes. The synced tables' hash covers foreign keys, so released apps
stop syncing with a server that has it until they update: deploy it
together with a release of Learn.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Tizian 「ティツィアーン」 <raisondetredev@pm.me>
2026-11-11 12:40:23 +01:00
..
bin chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
config build(deploy)!: the instance's compose file is compose.learn.yaml, beside compose.train.yaml and compose.todo.yaml 2026-11-02 23:12:15 +01:00
data chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
dictionaries chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
lib refactor(learn): own rows name shared content by its ID alone 2026-11-11 12:40:23 +01:00
migrations refactor(learn): own rows name shared content by its ID alone 2026-11-11 12:40:23 +01:00
test refactor(learn): own rows name shared content by its ID alone 2026-11-11 12:40:23 +01:00
.gitignore chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
analysis_options.yaml build: one workspace for every app and the packages they share 2026-10-22 18:11:03 +02:00
CHANGELOG.md chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
dart_test.yaml chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
Dockerfile fix(deploy): the images drop third_party/'s overrides instead of copying it 2026-11-04 13:43:46 +01:00
LICENSE chore: Halkyon Learn moves into apps/learn 2026-10-22 18:11:02 +02:00
pubspec.yaml fix(learn): the instance asks LanguageTool for its languages again a minute after it didn't answer 2026-10-26 16:21:52 +01:00
README.md build(deploy)!: the instance's compose file is compose.learn.yaml, beside compose.train.yaml and compose.todo.yaml 2026-11-02 23:12:15 +01:00

halkyon_learn_server

This is the starting point for your Serverpod server.

Start your server by running:

cd halkyon_learn
serverpod start

When you are finished, you can shut down the running server with Q.

Dictionaries

The dictionaries people look words up in (JMdict, …) are reference data the server holds for everyone. They come as dictionary files, which halkyon_learn_curator makes from the sources' downloads (its README says how). At every start, and every hour at seven past, the server loads the files in its dictionaries folder that changed since they were last loaded (by their SHA-256): in the background, a source at a time, and all at once, so searches meanwhile still find the old entries. A file it can't read is logged and left out. A dictionary whose file is gone is unloaded, once every file in the folder could be read (the curator removes the files of languages an instance no longer keeps). DICTIONARIES_DIR names another folder.

EDRDG's licence asks a server showing its dictionaries to update them at least once a month:

  • An instance keeps them current by itself: the dictionaries service in deploy/compose.learn.yaml (the curator, in a container of its own) downloads EDRDG's files of the day once its files are a week old, converts them, and replaces them in a volume the server reads (curate update, the curator's README); the server takes them up within the hour. To supply files yourself instead, remove the service and mount deploy/dictionaries/ into the server (compose.learn.yaml says how), and update them monthly.
  • Development: dart run halkyon_learn_curator:curate update --into ../halkyon_learn_server/dictionaries in the curator's folder (about a minute); the server takes them up within the hour.

JMdict in English, with the JLPT levels and Tatoeba's example sentences, is about 20 MB gzipped and takes the server half a minute to load (219,000 entries). KANJIDIC2 in English, with the JLPT levels of the kanji and the kana, is a dictionary of characters beside it (kanjidic-en.jsonl.gz, 0.7 MB, 12,800 entries): the app's Kanji & kana tab. Their licences (CC BY-SA 4.0; CC BY 2.0 FR for the sentences; CC BY for the JLPT lists) ask for attribution: the app lists the credits each file carries.

Licence

Copyright (C) 2026 The Halkyon Learn contributors

Licensed under the Server Side Public License, version 1 (SSPL-1.0); the full text is in LICENSE.