Data layer: Backend behind the seam
Los mantenedores suelen responder en 1 día
Nadie ha tomado este issue todavía.
Evaluación
- Dificultad
- 5/5
- Tiempo estimado
- Más de una semana
- Aptitud para principiantes
- 28/100
- Tipo de issue
- Nueva funcionalidad
- Claridad
- Bastante claro
- Estado de actividad
- Activo
- Stack tecnológico
- csharp, typescript
Línea de trabajo
Start with #584's C# repository and the existing seam, then read the suite's undo group and EditHistory for how undo and redo are currently exercised. The proposal leaves key design questions open, including Discard behavior and milestone scope. Done means a session runs on the C# repository behind a flag with undo and redo, while today's storage remains the default.
Escrito por el modelo de indexación a partir del texto del issue.
Descripción
This is a post-Phase-5 proposal, not a set of decisions. It's a starting point to discuss and change.
WebView sees: Nothing. Releasable because: the C# repository runs only behind a flag; today's storage stays the default.
Proposed direction
The CRDT store would keep no drafts: every edit persists. Save would go away. Undo and redo would use the CRDT's own history. Discard would stay in some form, with bookkeeping outside the CRDT if needed.
Scope
Plug #584's C# repository in behind the seam. Whether this is Phase 7 of the Host-owned data layer milestone or opens the next milestone is open (below).
The C# repository can be built against the suite's required group (#591, in #597) with none of these settled. They matter once a session runs on it.
⚠️ The revert in the first row below is the one piece of C# design here. Undo, redo, and Discard all build on it, so it is worth settling early (though it doesn't block #591).
| Piece | What it takes | Where it stands | Side |
|---|---|---|---|
| Undo and redo through the CRDT's history | Revert a given commit by writing a new one, keeping teammates' later changes to the same records. The seam would gain undo and redo calls that reply with a change set, and the suite's undo group, which tests apply's inverse, would move to them |
#584's Apply returns no inverse, and none would be needed. FW Lite has no undo over Harmony to reuse. Harmony drops a second create under an id already stored, even a deleted one (#597), so reverting a delete has to undelete the record, not create it again; reverting a project delete has to undelete every record the delete took with it |
C# for the revert; TS for EditHistory and the suite |
| Deletes that cascade | A ChangeBuffer guard that refuses deleting an analysis while a kept link names it |
Commands delete an analysis only once nothing links to it (reclaimIfUnlinked), so today only a test produces the case. Behind the seam, C#'s cascade would be a change no change set reports. #597 adds a second cascade: deleting an interlinear project deletes its records, and a record created in a deleted project lands deleted |
TS |
Session open and discards without exportAll (#588) |
Read the scope record and a query instead of the whole draft | Matters once the backend is in C#, where the whole draft would cross PAPI. #597's readAll command is that read, for the conformance suite |
TS |
| Record timestamps | Keep what a put carries, as DraftRepository does, or stamp the commit time, as #584 does |
Undo would no longer need a put's timestamps; imports, and Save As if it survives, still rely on them. The suite's required group leaves timestamps out of its comparisons, and a DraftRepository test pins today's behavior |
Either: a small C# change, or TS changes to what relies on it |
| Scope record (boundaries, languages) | Map it onto #597's InterlinearProject entity, or stay in today's storage |
#597's entity carries the name, description, target project, analysis languages (added and removed by name), and one segmentation delta with no split by book. CSharpRepository still refuses a change set carrying the scope record |
TS for the mapping; C# if the entity's shape changes |
Done when: a session runs on the C# repository behind a flag, undo and redo included, with today's storage still the default.
Before the C# repository becomes the default
| Question | Proposal | Why it waits |
|---|---|---|
| Discard | Revert this client's commits since a point, newest first, with their commit ids kept in host storage outside the CRDT | Needs the revert above, and the Discard point (open questions) |
| Save, Save As, the unsaved mark | Save would go. The unsaved mark could become "commits since the Discard point" | Follows from Discard; Save As is open |
| New | Would create the interlinear project at once (#597's createProject), since apply refuses a project its source's database doesn't store and there is no draft to hold work before it exists |
Follows from dropping drafts |
Later
| Question | Options | Why it waits | When |
|---|---|---|---|
| Re-anchor passes | Stored as commits, or derived on read | Only matters once two clients sync the same store | Before sync ships |
| Shards for projects | Likely never needed | Only if drafts go away before the CRDT store replaces them | If at all |
| When data stops being disposable | Pick the moment on purpose | Once a shared, synced store ships, schema changes need migrations and the change-set shape gets much harder to change | Before the first synced store ships |
Open questions
- Where does Discard rewind to: session open, a marker the user sets, or the last sync? Does its list of commits survive a restart?
- Does Save As survive, as copying a project to a new one, or go with Save? A copy would mint new record ids, since #597 refuses a put under an id stored in another project.
- Should this be Phase 7 of the Host-owned data layer milestone, or open the next milestone? Its done-when goes past the current milestone's.
- Lenguaje dominante
- TypeScript
- Estrellas
- 2
- Forks
- 0
- Merge medio
- 1 d 6 h
- PR fusionados (30 d)
- 88
Preparar el entorno
Este proyecto no incluye contenedor de desarrollo, Dockerfile ni guía de contribución, así que la configuración corre por tu cuenta: empieza por su README y consulta nuestra guía para la primera contribución para los pasos generales.
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Más de sillsdev/interlinearizer-extension
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
sillsdev/interlinearizer-extension#548 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 82/100
sillsdev/interlinearizer-extension#543 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 86/100
sillsdev/interlinearizer-extension#537 ·
Los mantenedores suelen responder en 1 día
-
bug
Dificultad 2/5 1-3 horas Aptitud para principiantes 76/100
sillsdev/interlinearizer-extension#515 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 4/5 3-5 días Aptitud para principiantes 25/100
sillsdev/interlinearizer-extension#593 ·
Los mantenedores suelen responder en 1 día
Todos los issues de sillsdev/interlinearizer-extension
Issues similares
-
Report: &TV InternationalAbiertocheck:passed streams:remove
Dificultad 1/5 Menos de una hora Aptitud para principiantes 67/100
Los mantenedores suelen responder en 1 día
-
bug priority-p1
Dificultad 2/5 1-3 horas Aptitud para principiantes 82/100
Los mantenedores suelen responder en 1 día
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 88/100
-
beginner blocked
Dificultad 2/5 1-3 horas Aptitud para principiantes 82/100
hiero-ledger/hiero-sdk-js#4670 ·
Los mantenedores suelen responder en 1 día
-
[Exploratory] Expense - 2 selected drop-down is shown twice when selecting hold from more option.AbiertoMonthly Not a priority
Dificultad 2/5 1-3 horas Aptitud para principiantes 65/100
Expensify/App#106080 · 1 comentario ·
Los mantenedores suelen responder en 1 día