docs: running.md said the azimuth level was still owed, three weeks after it landed - #61
Merged
Merged
Conversation
…fter it landed The dated account of the first lane run ends with "predicting the shift ... is a level of its own", which was true on 2026-08-14 and stopped being true on 2026-08-17 when fusion/azimuth.py landed. README.md meanwhile reports level 3 complete with the shift compensated before matching, so the two pages told a reader opposite things about what the chain does, and running.md is the one the README sends them to for real runs. The measurements stay exactly as recorded — the correction is only readable against them — and a paragraph after them says the level was built, what it recovers on this same scene (2 matches to 5 of 6 hulls), that the tolerance did not move, and where the approximation and the remaining debt are written down.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two pages told a reader opposite things about what the chain does.
README.md:51docs/running.md:162The code settles it:
src/darkvessel/fusion/azimuth.pylanded 2026-08-17 in1e62deb, is imported bypipeline.py:27andfusion/match.py:24, writesazimuth_shift_monto every classified detection, and is held by 13 tests intests/test_azimuth.py. The decision isdocs/decisions.mdunder 2026-08-16.running.md's sentence was true when it was written — the section is headedWhat the first run on the lane showed, 2026-08-14— and stopped being true three days later.docs/failures.mdsurvives this on its own because it is a chronological log: its 2026-08-14 entry says "recorded, and nothing else" and its 2026-08-16 entry records the correction.running.mdis not a log, and it is the page the README sends a reader to for real runs.The fix is additive. 10 lines, nothing removed, no measurement touched: the 341–632 m offsets, the implied
R/Vof 115 s and the table of six vessels stay exactly as recorded, because the correction is only readable against them. The new paragraph says the level was built, what it recovers on this same scene (matches from 2 to 5 of the 6 hulls in frame), that the tolerance stayed at 200 m — which is the point of moving the declaration instead of widening the radius — and points at the entry that carries the approximation (incidence angle declared at mid-swath, a fifth of the correction) and the debt still owed (the per-pixelangleband the export should record).make lintclean,make test572 passed / 1 skipped — unchanged.