You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
The "Is this the end of the wall?" prompt never appeared on a 19 ft walk #29
In field test run 1 (PR #23), the tester walked about 19 ft right of the meter and back. "Is this the right end of the wall?" never appeared, so "Can't get there" was the only way to finish a side, and that tap then hit #24.
Why. The end prompt appears only once reach covers 6.1 m (about 20 ft) without a gap (GuidancePlanner.swift#L28-L31, L133). A 6-inch cell counts only when all six sample rows have each been seen from two positions: the wall at 0, 3.25 and 6.5 ft up, and the ground at 0, 2 and 4 ft out. Replaying this scan's photos through the coverage rules:
the ground 4 ft out was never seen twice past 5.5 ft right of the meter;
the wall's top row, 6.5 ft up, was never seen twice past 10.5 ft.
Nothing on screen asked for either row. This overlaps known B-06 in the verification bug triage (an end closer than 20 ft can't be marked).
Possible fixes (your call):
while walking, prompt for whichever row is holding reach back (the top of the wall, the far ground);
let the homeowner mark an end wherever they stand, and record it as a real end, not "unexplored";
relax which rows reach needs.
The full run for context (1:40, Control Center and the share sheet cut out):
Addressed in 3c7b83a on t3/ios-mvf (merged in 75dea41). It lets you end the wall anywhere instead of making the prompt appear sooner.
What changed
During the walk, a "Wall ends here" button now sits next to "Mark something". It ends the side the card is asking about where you stand (the dashed line on the wall strip), then asks "It turns a corner / Something blocks it". "Something blocks it" records a real limit.
"Is this the right end of the wall?" now comes once you've walked about 20 ft (6.1 m) on that side, measured from where photos were taken, not from gap-free coverage. A hole near the meter no longer holds it back.
What's left
On a 19 ft walk the prompt itself still won't appear, because it needs 20 ft walked. The button is how you end a shorter wall.
The walk still doesn't ask for the rows that held coverage back on run 1 (the wall 6.5 ft up, the ground 4 ft out). Those are left to the aim prompts and the gap check after the walk.
To confirm on the first TestFlight build with 3c7b83a
End the left side first: the button ends whichever side the card is asking about, which is left until the left end is set. Then walk about 15 ft right of the meter. "Wall ends here" should be showing, with a dashed line under you on the strip. Tap it, answer the question, and check the right end is about 15 ft out. On a wall longer than 20 ft, keep walking instead: "Is this the right end of the wall?" should appear once you pass about 20 ft, even if the strip still shows gaps.
Field test run 2 (build 3.1) hit the other half of this: the mock wall ends at about 17 ft, so the 20 ft end prompt couldn't come. Only "Can't get there" was offered, which is B-06; the tester said "can't get there because I'm at this wall". That tap then put the right end 6 in from the meter (#24). A "Wall ends here" available whenever the end is in view would have avoided both.
Run 1, no end prompt on a 19 ft walk: by design. The prompt still needs about 20 ft (6.1 m) walked. It now counts the walked distance, so a hole near the meter no longer holds it back. On shorter walls, use the button.
Next field test (a build with 3c7b83a and #48): at the 17 ft end, check "Wall ends here" is showing with the dashed line under you, even during a tilt prompt. Tap it, answer "Something blocks it", and check the right end is about 17 ft out.
Fixed on t3/ios-mvf in 3c7b83a, merged at 75dea41 (B-06). Build d02cc11 or later for runs 2 and 3; it is green (CI run 36279804539).
"Wall ends here" is offered whenever you're walking a side that has no end, including during tilt prompts. It ends the wall at the phone's position (the dashed line on the strip), then asks what's there.
The "Is this the end of the wall?" prompt now comes once you've walked 6.1 m (20 ft) on that side. Before, it waited for 6.1 m of unbroken coverage, which run 1 never reached.
5c8c739: "Wall ends here" at the reticle refuses a point more than 6 m along the wall from the phone. A nearly parallel aim could otherwise put the end hundreds of metres away.
Build 4.1 field test (TestFlight 4.1, beta45017ba, iPhone 14 without LiDAR, three runs on 2026-09-26): looks fixed.
Run 1: "What's at the left end?" and "What's at the right end?" appeared at both physical ends (1:14, 2:47).
Run 2: "Is this the right end of the wall?" appeared on its own at the right end (3:15). The tester: "Oh shoot, it knows."
Run 2 noted that "What's at the left end?" appeared only after "Wall ends here" was tapped (1:45.5). In code, the end question is set only by markWallEnd and endWallHere, so it's worth checking whether run 1's questions were really automatic.
What happens after "It turns a corner" is now #70. Analyses: PR #60 (run 1, run 2).
Activity
AidenJohnston commentedon Sep 26, 2026
Addressed in 3c7b83a on
t3/ios-mvf(merged in 75dea41). It lets you end the wall anywhere instead of making the prompt appear sooner.What changed
What's left
To confirm on the first TestFlight build with 3c7b83a
End the left side first: the button ends whichever side the card is asking about, which is left until the left end is set. Then walk about 15 ft right of the meter. "Wall ends here" should be showing, with a dashed line under you on the strip. Tap it, answer the question, and check the right end is about 15 ft out. On a wall longer than 20 ft, keep walking instead: "Is this the right end of the wall?" should appear once you pass about 20 ft, even if the strip still shows gaps.
AidenJohnston commentedon Sep 26, 2026
Field test run 2 (build 3.1) hit the other half of this: the mock wall ends at about 17 ft, so the 20 ft end prompt couldn't come. Only "Can't get there" was offered, which is B-06; the tester said "can't get there because I'm at this wall". That tap then put the right end 6 in from the meter (#24). A "Wall ends here" available whenever the end is in view would have avoided both.
AidenJohnston commentedon Sep 26, 2026
Checked against
t3/ios-mvf(3c7b83a). Build 3.1 came from e9e405d, which doesn't have 3c7b83a.Next field test (a build with 3c7b83a and #48): at the 17 ft end, check "Wall ends here" is showing with the dashed line under you, even during a tilt prompt. Tap it, answer "Something blocks it", and check the right end is about 17 ft out.
SamGu-NRX commentedon Sep 27, 2026
Fixed on
t3/ios-mvfin 3c7b83a, merged at 75dea41 (B-06). Build d02cc11 or later for runs 2 and 3; it is green (CI run 36279804539).AidenJohnston commentedon Sep 27, 2026
Build 4.1 field test (TestFlight 4.1,
beta45017ba, iPhone 14 without LiDAR, three runs on 2026-09-26): looks fixed.markWallEndandendWallHere, so it's worth checking whether run 1's questions were really automatic.What happens after "It turns a corner" is now #70. Analyses: PR #60 (run 1, run 2).
WS-03 orchestrator: stages, retries, sweeper, early workers (#29)