A Duration is an absolute span, so stepping a local DateTime by
Duration(days: 1) moves 23 or 25 hours across a DST boundary and lands on
the wrong wall-clock day. Under Europe/Warsaw, spring-forward 2026-03-29:
lastCompleteDay(2026-03-30) -> 2026-03-28 23:00, the day before
yesterday, not yesterday
weeksAgo: 1 from 2026-04-01 -> 2026-03-24 23:00, a Tuesday, so the
wrong ISO week entirely
a 03-27..03-31 period -> emits four day keys, silently dropping
2026-03-31
The Python half computes all three correctly, because timedelta on a date
has no clock to drift against. So for about two weeks a year the phone and
the PC would report different averages for the same log -- exactly the kind
of device-local divergence the shared-source-of-truth rule exists to stop.
Passing out-of-range components to DateTime's constructor normalizes on the
calendar (day 0 is the previous month's last day), which is what timedelta
already does. The same fix goes to day_status_service's streak walker: it
had the identical pattern and predates this feature, so a streak that
spanned the clock change could break when it should have held.
None of the existing tests could see any of this -- every fixture date sits
in a DST-free span -- so the regression tests use dates that straddle
2026-03-29 and assert on the calendar day rather than the DateTime, which
keeps them meaningful whatever zone the runner is in. Verified they fail
against the old arithmetic before fixing it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DjJqpcEsaV5WRhPXyxCThd
The calendar already said which individual days went over budget, but not
whether the week as a whole did. A run of small overages reads as four
harmless yellow squares; averaged, it is the thing that actually moves the
needle.
Three rules are load-bearing, and are documented in CLAUDE.md because getting
any of them wrong silently changes what "under budget" means:
- The denominator is logged days, not elapsed days. An unlogged day is not a
zero-kcal day, and averaging gaps in as zeros would make a badly-tracked
week look like the best week of the year.
- The yardstick is the mean of the per-day budgets over those same days,
resolved through BudgetSchedule -- never today's budget, which would
retroactively reclassify past weeks on every budget edit.
- Today is excluded. A half-logged today drags the mean far enough to flip
the band, so every period ends at yesterday.
Band boundaries reuse the calendar's own OVER_BUDGET_YELLOW_CEILING by
import rather than by copy, so a period and its days can never disagree
about what "over" means; that constant loses its underscore to say so.
The MCP get_averages tool returns the average and the band but drops the
average budget, keeping the module's existing invariant that the budget
number never leaves this machine. The CLI, the gate and the phone all show
both, since it is hidden from the network, not from the user.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DjJqpcEsaV5WRhPXyxCThd
The largest consumer moves last on purpose: diet_guard was 71% of all sync
traffic, and both its surfaces (the PC timer and the phone app) share one
prefix, so they move together.
Firebase becomes the primary backend when it is set up, with GitHub kept as a
mirror so a device that has not moved yet still converges. An unconfigured
device keeps syncing exactly as before, which is also the rollback.
Revision tracking is new here. The PC hand-rolled its own pull-merge-push
instead of calling sync_log, so it had none: every tick re-downloaded every
peer's whole food log -- hundreds of KB, 96 times a day -- and re-pushed its
own regardless of change. It now reads one small revision map first, skips any
peer whose hash is unchanged, and suppresses a no-op push. Every sub-sync
(log, budget, food bank, manual bank) takes the client as a parameter, so one
swap covers all four.
The app could not have reached Firebase at all: nothing called saveAccount(),
so loadAccount() always returned null and openFirebase() always returned null.
Settings now has a Firebase section that stores the account in the OS keystore
and signs in immediately, so a typo surfaces on the screen where it can be
fixed rather than as a silently failed background sync hours later. The email
and password never touch the repo -- it is public.
Platform reaches now degrade instead of failing: an unavailable keystore or
application-support directory costs one tick of extra traffic, where before it
would have taken sync down entirely. Caught by `on Object` rather than
`on Exception`, because an uninitialised binding raises an Error.
Verified end to end against the real database, not just in tests: a live PC
sync merged 57 days / 339 entries and put 221 KB of food log plus 97 food-bank
records into Firebase; a second unchanged tick issued zero pushes; and the
Dart client signed in and read back all 339 meals and 97 foods, then completed
a push / publish-revision / suppress-no-op round trip under a throwaway prefix.
diet_guard: 618 tests at 100% branch coverage.
diet_guard_app: 477 tests. Two assertions moved from 4 pushes to 5 -- the
extra one is this device's revision, published after the log so a peer can
never cache "seen rev X" against a log it never received.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SJ9sxEw3ymzXNjL7SHNpJc
Four asks, plus the fixes two review passes turned up.
Removals (all unused):
- "Repeat last meal" button, and the reward/temptation-bundling prompt it
was the sole trigger for. Replaced with a post-log progress card showing
kcal against budget, macros, and the adherence streak.
- The multi-item meal builder, on both the gate and the phone. The stored
`components` field stays: nothing writes it now, but historical composite
entries and the food bank still read it and it is part of the wire format.
- Meal photos entirely: 13 files, `imagePath`, the photo filter, the
wrapper's /blobs/ route, image_picker, and the CAMERA permission.
Reminder false alarms had three real causes, all fixed:
- Logging never reconciled notifications, so a reminder sat on screen up to
15 minutes after the meal was logged -- on the phone itself.
- The background check never pulled, so a PC-logged meal nagged forever
until the app was opened. Now mirrors _cli_gate._should_lock: cheap local
check, network only when a slot looks due, then re-check.
- Meals before 08:00 / after 22:00 were attributed to no slot at all. New
slot_for_log/slotForLog clamps them to the first/last slot, byte-identical
across Python and Dart. Deliberately separate from elapsed_slots, which
would otherwise make every slot fall due at 23:00.
Budget history, so lowering the budget stops rewriting the past:
- Forward-only (effective_from, kcal) entries; each day is judged against
the budget in force then. Verified live: setting 2000 left a 36-day
adherence streak and the whole calendar pixel-identical.
- Rides as hist:<date> fields on the existing budget record. crdt_sync
merges per-field over the union of field names and both devices push the
merged log, so a device that predates the feature relays them untouched --
no coordinated rollout, no second document.
- The seed guard tests emptiness, not file presence: _sync can write an
empty history after merging with a pre-feature peer, and a presence check
would then never seed again, silently reclassifying every past day.
One shared source of truth for every piece of data:
- Body weight now travels as its own `weight` CRDT field. Inside `value` it
was collateral damage of whole-map LWW -- any phone budget edit deleted it,
and the protein target with it.
- Both halves of the food bank sync. The curated half (foods added without
eating them) was the last device-local state. The derived half merges with
count as the clock, so LWW means max-count-wins; the log stays
authoritative for membership, because a CRDT union never shrinks and a
peer's stale copy would otherwise resurrect a deleted food permanently.
- Bank adapters honour tombstones, and Dart parses bank records defensively:
a cast failure there is an Error, not an Exception, so one malformed
remote record would have broken phone sync permanently.
_test_guard.py blocks any pathlib write to real user state while pytest is
in the process, whatever conftest is loaded. A throwaway test run outside
the package conftest destroyed a real GitHub token during this work; this
makes that class of accident impossible rather than merely discouraged.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The IndexedDB stores are the desktop device's primary copy of the food log and
the *only* copy of its photos -- photos never sync -- and their wrapper-mirror
fallback is what makes a wiped Chrome profile recoverable. All of that was
invisible to the suite: the files import `idb_browser.dart`, which cannot be
imported outside a web compile, so they were simply absent from the VM test
run rather than covered.
Splitting the store classes from the browser-factory `open…()` functions lets
them run against `idb_shim`'s in-memory backend on the plain VM, which is what
the flutter-desktop-to-chrome skill recommends the shim for. Twelve tests now
pin the round-trip, the mirror write, the recover-from-mirror path, and the
no-wrapper-running case (a plain browser tab, where the app must still work).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Completes the move: `linux/` is gone, replaced by the web build served by the
wrapper in a Chrome `--app` window. The icons and the .desktop installer move
out of `linux/` first -- they are not Flutter scaffolding and are still
needed; the entry now points at the packaged launcher and matches the
`--class=diet_guard_app` the wrapper passes, so the taskbar shows the app's
icon rather than a browser's.
install_arch.sh carries over the three traps each of which produced an
install that looked fine until launched: `dart build cli` rather than
`dart compile exe` (which refuses any package whose dependency graph has
native build hooks), `options=('!strip')` (makepkg would otherwise discard
the AOT snapshot, leaving a binary that runs as the bare Dart VM printing a
usage message), and no browser in `depends` (this machine's policy uninstalls
`chromium` seconds later, failing dependency resolution).
Also adds the frame-cadence instrument the performance claim is measured
with, armed by `--dart-define=DIET_GUARD_FRAME_STATS=1` and read with
`String.fromEnvironment` -- `bool.fromEnvironment` only accepts the literal
`true`, so `=1` would evaluate to false and a whole run would silently
produce no data. It reports frame-to-frame cadence rather than build/raster
durations, since work that fits the budget says nothing about the frame rate
actually achieved.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Flutter's Linux embedder manages ~20fps at 3840x2160 -- measured on a stock
`flutter create` counter app, so it is the toolkit rather than this code --
where the same Dart rendered by Chrome sustains ~144fps (evidence in
~/todo/docs/desktop-performance-findings.md). The desktop target therefore
becomes a Flutter *web* build; Android is untouched.
`dart:io` does not fail a web *compile* -- it resolves to a stub that throws
at runtime -- so every platform edge moves behind a conditional export rather
than being discovered later as a blank white window:
- storage: `document_store*.dart`, files on Android and IndexedDB in the
browser (mirrored to the desktop wrapper, which is what makes a wiped
Chrome profile recoverable). Document names stay the exact pre-web
filenames so an installed phone keeps reading its existing log.
- photos: `blob_store*.dart` plus `attached_image*.dart`, since `Image.file`
cannot resolve an IndexedDB key. The picker itself has web support; what
changed is the storage and display model. Camera *capture* is genuinely
lost on desktop -- and photos never crossed devices anyway, `imagePath`
being stripped before push.
- scheduling: `background_tasks*.dart`. A browser has no WorkManager
equivalent -- a closed tab runs nothing -- so the desktop runs an in-page
5-minute check while its window is open. That loses nothing on the PC,
where `diet-guard-gate.timer` already locks the screen on an unlogged
slot, which is strictly stronger than a notification.
- notifications: `notification_backend*.dart`, the browser Notifications API
in place of flutter_local_notifications.
- the GitHub token: `token_vault*.dart`. On web it is *not* held by the
browser at all -- flutter_secure_storage's web backend is WebCrypto-wrapped
localStorage with the wrap key beside the ciphertext, which is obfuscation,
not encryption at rest -- so the desktop wrapper owns it and proxies the
API (`github_client_factory*.dart`).
`Platform.isAndroid` itself throws on web, so every platform branch now tests
`kIsWeb` first; the battery-optimization row (permission_handler, no web
implementation) is hidden off Android.
The desktop pushes under device id `desktop`, not `phone`: sharing an id
would have made the two devices overwrite each other's pushed log on every
tick. The Python side enumerates device directories generically, so it needs
no change.
Tests move onto the same seams -- the file-backed stores keep the atomicity
and unreadable-file guarantees where they belong.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Applies Atomic Habits' two-minute rule: a FAB on the home screen repeats
the most recent logged meal in a single tap, mirroring _onLogMeal's
existing post-write sync sequence. An optional device-local reward
(label + URL, configured in Settings) surfaces after a repeat and opens
externally via url_launcher.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Both platforms gain a calendar tab/screen (green/yellow/red/not-logged
day coloring, logging/adherence streaks, year-to-date tally), and
tapping a day drills into that day's logged entries. This requires the
daily budget to become a normal, freely-editable value synced between
devices like the food log — replaces the previous chattr +i seal
(deliberately, per user decision) with plain last-edit-wins sync via a
new budget.json alongside food_log.json. The budget field defaults to
2200 kcal, starts read-only, and unlocks via an Edit/Save toggle on
both the PC gate's History tab and the phone app's Calendar screen.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dHac1iYQKTMZEtL7xDJvA
crdt_sync v0.3.0 adds canAccessRepo(), so the settings 'Test connection'
button uses the shared client and the local lib/services/github_client.dart
(+ its test) is deleted. flutter analyze clean, 299 tests pass. On-device
foreground check deferred: the active night curfew disables the app (a
phone-policy thing, not a code issue); build + install -r succeeded.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019KwUCPESvE5CjCtmUFz8xQ
Immediate push on log only helps when the device is online. If a meal is
logged while offline, the in-process auto-sync fails silently and the meal
waited for the next app lifecycle event. Now each log also enqueues a one-off
WorkManager task constrained to NetworkType.connected, so the log uploads as
soon as connectivity returns, without the app being reopened.
- background_sync_service.dart: backgroundSyncPush() loads SyncSettings and
runs runSync, returning WorkManager's success/retry flag (true when nothing
to push or the push succeeded; false to retry after a transient failure).
Unit-testable half; the dispatcher branch and the one-off registration are
integration-only (real background isolate) and coverage-ignored, matching
the existing periodic check.
- log_meal_screen.dart: _enqueueSyncBackstop() registers the task (replace
policy coalesces a burst of logs; linear backoff retries) from both log
paths. No-op off mobile.
- background_check_service.dart: dispatcher routes the new task name.
- log_storage_service.dart: writeLog is now atomic (temp file + rename) — the
background push's runSync can write the log in a separate isolate while the
foreground app also writes, so a torn read must be impossible.
Verified on device (BL9000EEA0000102): airplane mode on -> log -> background;
airplane mode off -> the phone's remote food_log.json commit advanced within
~15s, app never reopened (also confirms the token loads in the background
isolate).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FtSmzACzJSGyEatgK9exkd
Replace the bespoke GitHub-Gist sync with the shared crdt-sync /
crdt_sync_dart libraries so diet-guard uses the same CRDT transport as
the other apps.
- Python: delete _sync_github.py (+ its test); rewrite _sync.py,
_cli_sync.py, _sync_merge.py against crdt-sync. Add the crdt-sync git
dependency (pyproject + pylint deps) and S107 to the test ignore list.
- Dart: rewrite app sync_service.dart / sync_merge.dart against
crdt_sync_dart (new git dep + crypto); update log_meal and settings
screens and their tests.
Verification: pytest 347 passed at 100% coverage; flutter analyze clean;
flutter test 306 passed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017RRCdR4D8SaVrN2xraMQxM
gatelock and diet-guard-sync were folded into shared monorepos:
gatelock -> github.com/kuhyx/utils (subdirectory), diet-guard-sync ->
github.com/kuhyx/syncs (subdirectory). Update the gatelock pip URL,
SYNC_REPO_NAME, the devices-dir path prefix (PC and phone), and add a
persisted-value migration so an already-configured phone picks up
'syncs' on next load.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Work-in-progress feature set accumulated ahead of the log-meal compact
layout change: a food bank browser, a shared entry-edit screen, an
app-wide settings service, and a substantially reworked history screen
with filtering/sorting.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018UorgLvWJ4huH55tmXoUAZ
M3 – GitHub OAuth device flow: replace PAT-paste with a guided "Connect
GitHub" button that runs the device-code flow; tapping with no client id
now opens a setup dialog (instructions + inline paste field) rather than
a buried inline hint. Bakes in the app's own OAuth App client id so fresh
installs work with zero manual config. Auto-syncs immediately after
connect. Verified end-to-end on the real phone: OAuth flow → token saved
→ PC's 48-entry log merged in (confirmed via food-bank vs manual source
labels in History).
M4 – Background meal-slot notifications: WorkManager periodic task (15 min
floor) checks for overdue slots and posts/cancels notifications via
flutter_local_notifications. New permissions: POST_NOTIFICATIONS,
WAKE_LOCK, RECEIVE_BOOT_COMPLETED, REQUEST_IGNORE_BATTERY_OPTIMIZATIONS,
INTERNET (was missing — latent sync bug). "Disable battery optimization"
button in Settings. Verified on real phone: WorkManager registered, forced
run posted a real notification ("Meal not logged / You haven't logged your
16:00 meal yet."), isolated to background path (only caller is the
WorkManager dispatcher, not any foreground lifecycle hook).
AGP9 release crash fix: AGP 9 defaults isMinifyEnabled/isShrinkResources
to true for release even with no proguard config; R8 stripped
WorkDatabase_Impl's reflection-only constructor, crashing every launch
with NoSuchMethodException. Explicitly disabled both flags in
build.gradle.kts. Verified via dexdump (constructor present) and on-device
launch (no crash). Proper R8 keep rules are the long-term fix; tracked.
177 tests, flutter analyze clean.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SWPUBzE24Ls9i9GMRwXnnn
Ports github_client.dart and sync_settings.dart from ~/todo (PAT-paste
instead of OAuth device flow), and writes a new (non-CRDT) sync_merge.dart
and sync_service.dart matching diet_guard's Python _sync_merge.py/_sync.py
algorithm exactly. Adds a settings screen for the PAT plus manual "Sync
now", and wires lifecycle-triggered auto-sync (launch + resumed/paused)
into the main logging screen, silent on failure per plan decision 4.
Also adds Linux desktop platform scaffolding so this and future UI changes
can be visually verified without a connected phone.
Verified end-to-end against the real kuhyx/diet-guard-sync GitHub API on a
Linux desktop build: Test connection and Sync now both round-trip to GitHub
and surface real auth errors correctly via SnackBar.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RH2BHCKbDTiYJUMG3rb9nq
Milestone 2 of the diet-app-as-wise-balloon plan, plus feedback from
manually testing it on-device:
- PhotoAttachService wraps image_picker and copies the picked photo into
<app documents>/images/<uuid>.<ext>, so the file survives after the
picker's own (possibly cache-cleared) temp copy is gone. Phone-local
only, per the sync plan: imagePath is never synced.
- PhotoAttachField is a shared attach/preview/remove widget, used
identically by both the single-item and composite-meal logging
screens, so logging a multi-item meal can now carry a photo too.
- PhotoViewerScreen gives a full-screen, pinch-to-zoom view of an
attached photo -- the 64x64 inline thumbnail was too small to
actually check the photo.
- HistoryScreen lists every logged entry across all days, newest first,
with a thumbnail when one is attached. There was previously no way to
confirm what got logged (or whether a photo actually attached) short
of inspecting food_log.json directly.
Verified on a physical device (BL9000): built, installed, and the user
confirmed the photo-attach flow logs a real entry with a real photo,
visible afterward in the new history list. 88 Flutter tests passing,
`flutter analyze` clean against very_good_analysis.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FU3f5KQ1GHXsbbSecfVEyF
Milestone 1 of the diet-app-as-wise-balloon plan: a phone-native way to
log meals away from the PC, sharing the exact on-disk JSON shape
diet_guard already uses (same field names, no translation layer).
- lib/models/: 1:1 Dart mirrors of the Python dataclasses (Nutrition,
FoodEntry, MealItem, FoodBankRecord, Slot), including the per-100g/
amount-eaten portion scaling that matches _resolve.resolve_nutrition's
semantics exactly.
- lib/services/log_storage_service.dart: plain-JSON persistence to
food_log.json's exact shape (no sqflite -- the canonical format
already is this JSON).
- lib/services/foodbank_service.dart: ports _foodbank.py's upsert/fuzzy
search logic for autocomplete.
- lib/screens/: log_meal_screen.dart (single-item logging) and
meal_builder_screen.dart (composite multi-item meals, logging full
per-component macros via the new components field).
Verified end-to-end on a physical device (BL9000): built, installed,
logged a real meal through the UI. 77 Flutter tests passing, `flutter
analyze` clean against very_good_analysis.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FU3f5KQ1GHXsbbSecfVEyF
