Conversation
A white block on this panel can make a phone's exposure stop down until the code stops resolving, and the backlight cannot fix it: it moves the whole screen rather than the code's contrast against the room. The light half of a QR is drawn in one of five greys now, white down to near-black and round again, starting at white. An address's QR carries a sun beside the code to cycle it - drawn from rectangles, since both faces here are ASCII - whose disc takes the current shade, so the glyph dims with the code. Verified by: host self-test (one press moves the light field to one new colour and nothing else, the ring closes, and the highlight travels to the sun - confirmed failing when the sun ignores the cursor), 52 Python tests, QR smoke, ASan/UBSan self-test, firmware at 305520 bytes. Not flashed; no camera has judged it.
The shade control could not reach the animated codes: up and down were stepping BBQr frames. Manual stepping goes - a reader collects the parts across repeats, so a held frame never completes a scan - and the cadence slows to 1000ms. The Compact SeedQR's views are not frames, so there the sun becomes one more stop past the last tile. draw_qr_nav goes too; nothing had drawn it. Verified by: host self-test (the highlight reaches the sun on all five QR renderers, confirmed failing when one ignores the cursor), 52 Python tests, QR smoke, ASan/UBSan self-test, firmware at 305536 bytes. Not flashed.
A taproot address does not fit beside its code and falls back to the plain QR, which draws the same chrome as every other one - so it was drawing a sun that answered nothing, on the value hardest to scan. It now takes the same two keys as the rest. Two things the chrome left behind: draw_qr_control's position and direction have been frozen since the forward control went, so it goes back to being the one control it draws; and the gap between the sun and the code, a bare 3, is named and pinned. A code is never wider than the display is tall, so the sun's box can never reach the back arrow - a static assert says so, since two boxes overlapping would draw quietly rather than fail. Verified by: host self-test, 52 Python tests, QR smoke, ASan/UBSan self-test, firmware at 305584 bytes. The assert was confirmed to fire.
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
1 participant
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.
A white block on this panel can make a phone's exposure stop down until the code
stops resolving, and the backlight cannot fix it: it moves the whole screen
rather than the code's contrast against the room. Jade ships a five-step ramp for
the same reason, and on its ESP32 screen does not even default to white
(
main/gui.c:88).What changes
and round again, starting at white — so a code nobody touches is drawn exactly
as before. The dark modules never move: contrast is raised from one side only.
the code's left edge that cycles the shade. Up and down move between them, the
chord takes the highlighted one. The sun's disc is drawn in the current shade,
so the glyph dims with the code; it is built from rectangles because both fonts
here are ASCII and a symbol font is a dependency for one glyph.
repeats, so a frame held still can never complete a scan — the feature was
spending both keys on every animated screen to do nothing useful. Removing it
frees them for the chrome and takes
bbqr_step_partand the manual/auto splitout. The cadence slows from 700 ms to 1000 ms to pay for what is lost.
(the code, the region map, the tiles meant to be hand-copied), not frames of
one payload. The sun becomes one more stop past the last tile, leaving the view
under it standing.
draw_qr_navand its forward control are removed: nothing had drawn them sincethe address screen stopped handing off to a second screen.
Verified by
renderers, and one press moves the light field to one new colour and nothing
else, with the ring closing on itself
the cursor; the ramp made to wrap instead of report)
-Werror, clang ASan/UBSanself-test
mainNot verified
No camera has judged any of this. The branch is flashed and the screens have
been walked, but whether a dimmer code actually scans better — and whether
1000 ms is the right cadence now that a frame cannot be held — are questions only
a phone in a real room answers. Both numbers are guesses, as the 700 ms before
them was. If the answer comes back "white always wins", this should be reverted
rather than kept for tidiness.