| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The second real-watch round on the Forerunner 70 passed: codes from
otpauth:// links, the 4-digit PIN, touch scrolling and Aegis JSON import.
Version 1.0.0 ships the fixes of the 0.9.0 review.
- docs/STORE_LISTING.md: title, description and What's New for the
Garmin developer page, with measured character counts.
- docs/store/: hero (1440x720), cover (500x500) and seven round screen
images, built by tools/store_assets.py from the fr70 goldens.
- CHANGELOG 1.0.0, README status, TESTING round 2 results, YOUR_STEPS
marked done, fr70 about.png golden at 1.0.0.
make ci green on fr70.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
After a correct or new PIN, promoting pending entries and applying waiting
"Require PIN" changes ran in the event of the last PBKDF2 step; 16 waiting
changes tripped the watchdog on the Forerunner 70. ChangeQueue.LockJob now
does it in steps while the PIN screen shows "Checking...", saves what it did
to the results and refreshes the glance after Set PIN too.
The list title decrypted every waiting change at each start. Queue items
are now [action, box] (aad OTPW-Q-2 + action code), so they are counted
without decrypting; 0.9.0 items are still read. A replacement is checked
when it arrives and its material is sealed apart (qm.<n>). Large buffers
are wiped and UTF-8-checked in pieces (Jobs.Wiper, Bytes.utf8Prefix,
SealedBox.openKeep/sealKeep). "Phone changes" reads behind "Reading...",
and Confirm all / Reject all act only on the changes shown. Storage full
no longer stops the app at the PIN.
Found by the review of b8addc8/096b5a4 and a new make watchdog test; the
diff was reviewed again and its findings fixed. fr70: make lint, make test
(176), make watchdog (19 cases at two steps per event), make screenshots
pass; list_locked and list_reset goldens show "Results ready (menu)".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A second UI-test build (uitest-extra.jungle) covers the legacy PIN and
the forced new PIN, Change and Reset PIN, a queued Require PIN, UPC-A,
Code 39, the code error screen, the encoding progress, auto-lock on a
barcode, the light theme and the glance without a code; the main flow
adds the Steam and Library details and the menu's text screens.
Fixed from what it found: Entry IDs tripped the watchdog with 110
entries; auto-lock didn't close a locked barcode; a queued Require PIN
showed "Confirm in menu"; locked barcodes lacked the padlock; the PIN
keypad's bottom row, the Reset PIN question, text-screen bottoms, the
glance title and the light-theme warning colour.
Goldens approved on all seven watches; only the Forerunner 70 was
re-verified after the version bump.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
| |
The QR percentage counted phases and the display froze at 90 % once
encoding ended, so the first opening sat at 90 % while the slow stages
ran. Each 2D symbology now weights its phases by measured cost,
interpolates inside the long ones, and the display splits encode,
pack and runs; a cached code counts its runs from 0.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
The keypad has no OK key; the 4th digit submits, so a tap that also
arrives as START can no longer submit a partial PIN (the likely cause
of "Wrong PIN" before the code opened). PINs set by earlier versions
still unlock and are then replaced by a 4-digit one. Opening an entry
ignores a second request from the same tap. A Require PIN sent from
the phone while the watch is locked now applies at the next PIN
instead of waiting for Confirm all. The linter skips .claude.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
screenshot
- make watchdog (in make ci): a stress build encodes, packs, turns into
runs and draws the largest code of every 2D symbology at two steps per
event on the Forerunner 255S and 70; it fails on a watchdog trip, and
the last code left on the screen must decode (bitmap path on the 70).
- make memtest checks the glance (DIAG glance, budget 70 % of 64 KB).
- screenshots.sh launches the app again after the flow and compares the
glance with the favourite's code (glance_code.png on 7 watches); the
About goldens show 0.8.0.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With "Show the favourite code in the glance" on, the glance shows the
code of the first favourite TOTP or Steam entry without the PIN, with a
countdown bar. The glance can't open the vault, so the About scan seals
that entry under "g" (device key, aad OTPW-GL-1) after every change, and
removes it when the option is off or no entry qualifies; the glance also
reads the option itself. HKDF moves to its own module so the glance loads
only the crypto it needs (37 % of its 64 KB). Manage entry gains "Mark as
favourite (glance)" and "Remove favourite", queued for locked entries
like the other changes. Lines are inset on round screens so the
Instinct 3's curved glance corner doesn't cut them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
codes from runs
M6's step sizes were estimates, only exercised with small codes. With the
largest inputs the app accepts, every 2D encoder tripped the watchdog on
the Forerunner 255S (and the biggest steps were far over the 768 KB
watches' limit), so a large code added from the phone would have stopped
the app during settings processing at every start.
- QR: resumable segmentation, Reed-Solomon, interleaving, function
patterns and final mask; mask scoring on rows and columns packed 16
modules per word (same penalties, checked against the per-module rules).
- Reed-Solomon: resumable field tables and generator, encoder shifting
with native slice.
- Data Matrix, Aztec, PDF417: resumable encodation, stuffing, compaction,
placement and build; Data Matrix's grid as a byte array (90 KB -> 35 KB);
PDF417 refuses text that can't fit the screen before compacting.
- Cache format 2 with 16-bit sizes (wide PDF417 was cached wrong); packing
in steps; codes drawn as runs, and from a bitmap painted in steps when a
frame can't draw all runs.
- The job runner does two steps per tick on 768 KB watches;
gen_products --check verifies each product's watchdogCount fits that.
- CodeView renews full brightness (AMOLED) or the backlight (MIP) for up to
a minute from its single ticker; a tap restarts the minute.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
A parsed import item with its Entry took about 760 bytes as objects on the
Forerunner 255S; packed into one Codec ByteArray it takes about 110, and
the importer unpacks one item per store step (wiping the packed secret).
Aegis group names are resolved when unpacking, so no step repacks all
items. The JSON is parsed from the untrimmed text's bytes and the string
is dropped; an incomplete part is saved from those bytes. The form hash
is computed before the importer copies the text.
The peak of a 100-entry import on the 255S falls from 56 % to 45 %.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
forms end to end in the UI flow
34 Monkey C tests (154 per device): validation messages, idempotency, the
queue, the About scan, "Encrypt names" and the properties contract. The
UI-test build fills in Add OTP, Add code and Manage like the phone; the flow
checks the new TOTP (309 848 at the frozen time), decodes the Code 128 card
and confirms the queued change to the locked entry with the PIN.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
| |
screenshot
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
| |
also from screenshots
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
| |
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
| |
and screenshots
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
| |
entries in the UI flow
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
| |
tools/ui_flow.py clicks buttons and taps from the device files, waits for
DIAG shown lines and compares 10 screens per watch with goldens on four
test and three spot-check devices. Adds the fixture-wrap generator, UI
logic tests and driver tests; the launcher opens apps from their glance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
| |
200 byte-wise iterations in one event tripped the 120,000-bytecode
watchdog on the Forerunner 255S profile. Word-wise XOR makes PBKDF2
2.3x faster (22 ms per 1000 iterations in the simulator).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
| |
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
| |
34 new tests (65 per device): round trips per entry type, every tampered
byte, swapped records, the pending queue, the entry cap, a full store at
67 fill levels, the lockout schedule on a fake clock including restarts,
erase, PIN change and reset, auto-lock wiping, and v0 to v1 migration.
Adds an in-memory store with a size limit and storage benchmarks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
- gen_vectors.py: RFC/FIPS/NIST vectors recomputed with Python's stdlib
(openssl for AES) before use, plus seeded cases around SHA padding
boundaries, 2038 and 2100; generates test/data/Vectors.mc.
- 31 Monkey C tests (hashes, HMAC, PBKDF2, HKDF, AES, sealed records incl.
every-byte tamper detection, Base32 errors, 100 OTP and 12 Steam
vectors, clock, jobs) and a benchmark; pass on all four test devices.
- Screenshot tolerance 0.01 %: one changed character no longer passes.
- Golden images updated for version 0.2.0.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
| |
|
|
|
|
|
|
|
|
| |
The M0 placeholder app: AppBase entry point, home screen with name and
version, debug-only diagnostics (start-up time, memory) and a unit test
that runs in the headless simulator. Builds with strict type checking
and zero warnings for all 72 products; icons.jungle gives every product
a launcher icon of its exact size.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
|
|
- sim_screenshot.py finds the device drawing in the simulator window,
crops the screen and compares it with approved golden images.
- check_budgets.py reads key=value DIAG/HARNESS lines (the simulator
console escapes quotes, so JSON was dropped).
- ciq_test_report.py accepts SDK 9.2.0's real summary line
`PASSED (passed=N, failed=N, errors=N)`; a captured log is a fixture.
- gen_products.py checks all manifest products are watches with
API 5.0+ and at least 512 KB for apps.
- lint_project.py rule HDR001: every file carries the Author/Summary
header (JSON keys, PNG text chunks; generated package-lock exempt).
- make_icon.py --all draws one icon per launcher size in use.
- Golden screenshots for fr255s, venusq2, fr70 and fr965.
- 44 tests, warnings as errors, ruff and mypy --strict clean.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
