chore: ng-e2e-helpers suit alpha.14, et la mise a jour des paquets devient une tache

This commit is contained in:
Sylvain Duchesne
2026-09-21 10:52:07 +02:00
parent d0ca5c5d41
commit 4a4e6af337
9 changed files with 107 additions and 53 deletions
@@ -19,12 +19,18 @@
* scope still syncing reads `{ data: [], isPending: true, isSuccess: false }`.
*
* ── The scope is the SDK's, and translating it is EXPLICITLY TEMPORARY ─────
* `scope` is `@ng-org/orm`'s own `Scope` — `{ graphs?: string[] | string; subjects?:
* `scope` is `@ng-org/orm`'s own `Scope` — `{ graphs: string[] | string; subjects?:
* string[] }` — not the three strings this module took before. A `graphs` entry names
* ONE document's graph, `did:ng:i` (or `""`) discards the rest and means the whole local
* dataset, and `[]` or `undefined` returns no objects at all: all three are upstream's
* rules, reproduced by `normalizeScope` below.
*
* `graphs` is REQUIRED as of `@ng-org/orm@0.1.2-alpha.21`; it used to be optional. Only
* the spelling changed, never the meaning: a caller who wants "everything" still has to
* say `did:ng:i`, because on this axis an empty list is the EMPTY set and not the absent
* filter. `subjects` keeps the opposite convention - there `[]` means no filtering - so
* the two axes read `[]` in opposite directions and `normalizeScope` keeps them apart.
*
* Only the whole-dataset reading still has to be TRANSLATED, and the translation is the
* tell that the mechanism under it is about to stop being ours: it is served by
* enumerating this identity's own documents across its three stores