Sender-side offline delivery hardening.
- OfferOutbox: a keystore-backed (no plaintext) queue of lot ids to publish
later. Stores ids, not offers, so on flush each offer is rebuilt from the
lot's CURRENT state — an offline edit is respected, a deleted lot drops out.
- flushOutbox(): on reconnect, republishes queued lots and clears them (no-op
offline; drops ids whose lot is gone).
- Market screen: sharing offline parks the lots ('we'll share these when you're
connected'); opening the market online auto-flushes the queue first.
- Wired via DI + TaneApp(outbox); i18n en/es/pt.
Tests: 4 outbox + 3 flush + market UI, 20 green. Analyzer clean for new code.
Persistence + wiring the offers screen needs, per the product decisions:
- SocialSettings (keystore-backed, no plaintext): coarse area geohash +
configurable relay URLs, both empty by default (no bundled public relays —
don't leak metadata to third parties). Registered in DI.
- createOffersCubit(): opens a SocialSession on the first configured relay when
online, else a null-transport cubit that degrades gracefully. OffersCubit now
closes the owning session on dispose.
4 settings tests green; offers/social suites still green.
Third slice: bridge Block 1's root seed to the Block 2 transport foundation.
- SocialService: derives the secp256k1 Nostr identity from the SAME root seed
the recovery QR backs up (no extra backup), exposes npub/pubkey, and opens a
SocialSession = one shared connection carrying all three transports.
- injector.dart: registers SocialService in get_it. Local-first — derivation is
cheap + offline; no relay is contacted at startup. Replaces the old
"nodeId slice of the seed is the social pubkey" placeholder; keeps the CRDT
author id distinct (unifying it would rewrite authorship — a later decision).
- Test: deterministic identity from the seed hex, matches commons_core, rejects
malformed hex. 5 tests green; app_seeds analyzes clean (0 errors).
'Save a backup' now writes a sealed .tanemaki file (AES-256-GCM under a
key HKDF-derived from the root seed; format TANEBK v1, open and stable).
Restoring on the same identity is silent; a copy sealed by another
identity asks for the printed recovery code (TANE1 base32, typo-tolerant)
and adopts that seed — recovering your bank recovers you. Legacy plain
exports still restore. Settings gains a recovery sheet (QR + code PDF).
BackupBox/RecoveryCode live in commons_core, pure Dart, TDD.
The local half of the original Fase 2 (no network, no Offer entity yet):
- Lot sheet gains a 'do you share it?' choice (gift/swap first, sale last,
never the default); declaring plenty of abundance reveals it with a nudge.
- Lot lines and list tiles badge shared batches; an 'I share' filter and
a printable PDF catalog (paper bridge for fairs) round out the view.
- Catalog PDF embeds DejaVu (Latin-ext/Cyrillic/Greek); fonts are a
per-locale resource like the OCR packs.
Lower the bulk-digitization cliff with two more routes on top of the
already-landed CSV import and "save and add another":
- Photo-first drafts (capture now, catalogue later): burst-capture
photos (camera or multi-gallery) into unnamed draft varieties, shown
in a "to catalogue" tray, hidden from the main list until named.
Adds Variety.isDraft (schema), addDraftVariety/watchDrafts/nameDraft,
the triage sheet and the inventory banner.
- On-device OCR label suggestion (Tesseract, offline, no Google): a
"Suggest name from photo" button in the naming dialog behind a
LabelTextExtractor interface (Tesseract on Android/iOS, no-op
elsewhere). Reads the largest print via hOCR bounding boxes, drops
boilerplate/low-confidence noise, preprocesses (grayscale, contrast,
upscale) and sweeps rotations (0-315 deg) so tilted packets still
read. Bundles tessdata_fast eng+spa; validated on-device against real
packets. The photo is written to a temp file deleted immediately in a
finally block (the plugin needs a path) - a bounded, documented
exception to no-plaintext-at-rest.
This commit also carries the co-developed schema evolution v5 to v8 that
shares these files (organic flag, species viability years, crop
calendar, lot provenance/abundance/preservation format, condition
checks) plus their exports/migrations and i18n.
Tests: CSV/draft/OCR unit + widget + migration green in isolation.
Note: the full widget suite currently hangs (>10 min) - under investigation.
Interchange export/import for Phase 1 (data-model §7):
- JSON: canonical, versioned envelope (formatVersion 1) with all sync
metadata verbatim, photos embedded as base64, tombstones excluded.
Species are re-resolved on import by scientific name (catalog ids are
per-install). Reader tolerates unknown fields/enum values (§5.2) and
rejects newer format versions with a clear error.
- CSV: export-only spreadsheet flatten, one row per lot, RFC 4180
escaping, never any photo bytes.
- Import merges by UUIDv7 id in one transaction: insert-if-unknown
preserving original stamps, last-writer-wins by packed HLC for known
mutable rows, append-only movements; afterwards the local clock
receiveEvent()s the newest imported stamp so it never runs behind.
- Settings gains a Backup section (export CSV/JSON, import JSON with
confirmation); file dialogs behind a FileService interface backed by
file_picker (MIT).
- Tests: codec round-trip and tolerance, reconciler LWW, repository
export→import round-trip (fresh DB, idempotent re-import, newer-local
wins, clock monotonicity, species re-resolution), backup widget flows.
Add a small curated catalog of Iberian horticultural species and let a variety
be linked to it from the edit sheet.
- assets/catalog/species.json: 14 species with botanical family and ES/EN
common names (wikidata_qid/gbif_key deferred to the varilla enrichment).
- SpeciesRepository: idempotent seedBundled (is_bundled rows, keyed by
scientific name) + search by scientific/common name with a locale-best label.
Seeded on startup from DI.
- VarietyRepository.linkSpecies: sets species_id and prefills category from the
species' family when empty (never overwrites an existing category).
VarietyDetail now carries the scientific name.
- Edit sheet gains a live species-search field; the detail view shows the
scientific name (italic). i18n strings added (ES/EN).
Tests: catalog parse, idempotent seeding, search by scientific/common name,
linkSpecies prefill semantics, and a widget test for the autocomplete → link →
scientific-name-shown flow. Full suite: 32 passing, 0 skipped.
Stand up the Tanemaki monorepo and the first end-to-end vertical slice of
Block 1 (offline, encrypted inventory): add a seed → see it in a categorized,
searchable list → it persists → reopen and it's still there.
Architecture (state management like G1nkgo, adapted to Tane's reality):
- flutter_bloc (Cubit-first), but the encrypted Drift DB is the single source
of truth; cubits stream from repositories (no hydrated_bloc/Hive, which would
write plaintext at rest).
- get_it composition root; go_router; slang i18n (ES/EN, Weblate-friendly JSON).
Workspace & core:
- pub workspace: packages/commons_core (pure Dart) + apps/app_seeds (Flutter).
- commons_core primitives: UUIDv7 IdGen, Hybrid Logical Clock, Quantity value
type (+ plant-aware QuantityKind), IdentityService root-seed stub.
Data & security:
- Drift schemaVersion=1 with all 10 Block-1 tables + common CRDT columns
(HLC updated_at, last_author, tombstones); Movement append-only.
- SQLCipher via an injectable executor that refuses to open a plaintext DB;
DB key + root seed in the OS keystore (separate secrets).
- Exported drift_schema_v1.json + migration scaffold.
Tests (near-TDD; nothing merges without tests):
- commons_core units (24), Drift migration test, "no plaintext at rest"
security guard (runs where SQLCipher is present, skips otherwise),
repository, widget, full quick-add flow, file-reopen persistence, and a
no-hardcoded-strings i18n guard. GitLab CI: format + analyze + test + coverage.
Follow-on Block-1 stories (item detail/edit, species catalog, germination UI)
and all of Block 2 remain out of scope.
