docs: correct the drift, and stop linking to files that are not there (#18) - #37
Merged
Conversation
… parts Closes dni#14. src/ui/buttons.c has been pointing at docs/HARDWARE-TEST-CHECKLIST.md for a while; the file did not exist. It does now, in the format the issue asks for: numbered sections, each ending in a dated bench record naming the board and the firmware version get_info reported, and an explicit NOT YET BENCH-RUN in those words where nothing has been run -- so nothing counts as verified merely by having been written down. Fifteen sections. Nine carry a real record from 2026-08-17 on a classic LilyGo T-Display; six say NOT YET BENCH-RUN, and say why: the approval gesture and granting a wipe need a finger on the board, OTA needs the release signing seed, the watchdog firing needs code that would have to be removed again, release artefacts need a tag, and the S3 target builds but was not bench-run in this pass. test/hardware/bench.py runs everything that does not need a press, prints a pass/fail table and exits non-zero, so it can gate a release. It deliberately never grants a wipe and never approves an export: an automated test one press away from erasing a device is not a test worth having. It degrades to skips against firmware that predates a feature rather than reporting failures, so it is usable for finding out what an older build does. Two things it handles that cost real time to learn, and that are written down next to the code that handles them: opening the serial port drives DTR/RTS into the auto-reset circuit, so a script that opens per command reboots the device between every command and silently invalidates anything measuring uptime or persistence; and boot chatter shares the UART that carries the protocol on the classic board, so responses have to be matched by looking for a line starting with '{'. Verified: run against a board carrying every currently-open PR merged together -- 23 passed, 0 failed, 5 skipped, all five skips being the rows that need a human. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RpMeqfvv9KqifKGVM2GwA5
Closes dni#18. WHAT WAS WRONG The README said browser-side integration into lnurl-wallet "is a deliberate follow-up, not part of this repo yet", and listed "No lnurl-wallet integration yet" under known limitations. It exists: device.ts, deviceQueue.ts, deviceOrchestration.ts, DeviceContext.tsx and pages/Vault.tsx, with device.test.ts and deviceOrchestration.test.ts. Checked each file against dni/lnurl-wallet rather than taking the issue's word for it. sdkconfig.defaults opened by saying it had been authored without a working ESP-IDF install, and pointed at a README section called "Status: unverified by compilation" that no longer exists. CI builds both environments on every push and both have run on hardware. The useful half of that note -- that option names drift between IDF releases -- is kept, now next to the fact that platformio.ini pins the platform so it does not happen by surprise. The nvs_keys note said the partition was "now-unused". It is not in the repo at all, and the sentence stopped short of the thing a reader most needs: NVS encryption is off entirely, so a physical flash dump recovers every secret and physical possession is the protection model. Said plainly now. Reported build sizes were 30.1% RAM / 25.6% flash. A real build of this commit is 36.3% / 27.2%. Two claims that hardware has since contradicted: a "What's still unverified" paragraph listing display rendering, button timing, BLE and NVS persistence as open questions, and a line calling export_secret's approval gate "unverified on hardware". All of those have been exercised. That paragraph now defers to docs/HARDWARE-TEST-CHECKLIST.md, which carries dates, boards and firmware versions per area and marks anything unrun as NOT YET BENCH-RUN -- a file is a better authority for this than a prose paragraph that goes stale silently. LINKS THAT DID NOT RESOLVE Seven links pointed at ../luds/25.md or ../../luds/25.md, plus ../lnurl-wallet and ../lnurl-mint. Those only work inside a monorepo checkout; on GitHub they 404. LUD-25 is not a merged file anywhere -- it is an open PR, lnurl/luds#301 -- so the links now point at the proposal, and the two sibling projects at their own repositories. NOT CHANGED The web installer's github.com/dni/lnurl-vault source links. The issue lists them as drift, but they point at the canonical upstream and are correct as written. Deriving them from the Pages host would add JavaScript to a page whose job is flashing firmware, and would not survive a custom domain. Left alone deliberately rather than churned. Verified: 257/257 native assertions; the S3 target builds (the source of the corrected size figures); the LUD-25 PR and all five lnurl-wallet files confirmed to exist via the GitHub API. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RpMeqfvv9KqifKGVM2GwA5
TheCryptoDonkey
force-pushed
the
docs/fix-drift
branch
from
August 17, 2026 15:42
ca4dd3d to
45159b0
Compare
dni
approved these changes
Aug 17, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #18. Stacked on #36 — the README now points at
docs/HARDWARE-TEST-CHECKLIST.md, which that PR adds, so stacking keeps the link from dangling at any point. This PR's own diff is the last commit.What was wrong
The wallet integration exists. The README said it "is a deliberate follow-up, not part of this repo yet" and listed "No
lnurl-walletintegration yet" under known limitations.device.ts,deviceQueue.ts,deviceOrchestration.ts,DeviceContext.tsxandpages/Vault.tsxare all there, withdevice.test.tsanddeviceOrchestration.test.ts. I checked each file againstdni/lnurl-walletvia the API rather than taking the issue's word for it.sdkconfig.defaultsdisowned itself. It opened by saying it had been authored without a working ESP-IDF install, and pointed at a README section called "Status: unverified by compilation" that no longer exists. CI builds both environments on every push and both have run on hardware. The useful half of that note — option names drift between IDF releases — is kept, now sitting next to the fact thatplatformio.inipins the platform so it doesn't happen by surprise.The
nvs_keysnote undersold the consequence. It called the partition "now-unused"; it isn't in the repo at all. And it stopped short of what a reader most needs to know: NVS encryption is off entirely, so a physical flash dump recovers every secret and physical possession is the protection model. Said plainly now.Stale numbers. Reported sizes were 30.1% RAM / 25.6% flash. A real build of this commit is 36.3% / 27.2%.
Two claims hardware has since contradicted: a "What's still unverified" paragraph listing display rendering, button timing, BLE and NVS persistence as open questions, and a line calling
export_secret's approval gate "unverified on hardware". All exercised now. That paragraph defers to the checklist instead — a file with dates, boards and firmware versions per area, which marks anything unrun asNOT YET BENCH-RUN, is a better authority than a prose paragraph that goes stale silently.Links that did not resolve
Seven links pointed at
../luds/25.mdor../../luds/25.md, plus../lnurl-walletand../lnurl-mint. Those only work inside a monorepo checkout — on GitHub they 404.LUD-25 is not a merged file anywhere; it is an open PR, lnurl/luds#301. The links now point at the proposal, and the two sibling projects at their own repositories.
Not changed, deliberately
The installer's
github.com/dni/lnurl-vaultsource links. The issue lists them as drift, but they point at the canonical upstream and are correct as written. Deriving them from the Pages host would add JavaScript to a page whose job is flashing firmware, and would not survive a custom domain. Say the word if you'd rather they were fork-aware.Verification
lnurl-walletfiles confirmed to exist via the GitHub API