Consolidated follow-ups from the June/July sweeps (#73, #88, #97)

[espg]

🤖 from Claude

Consolidation of the three rolling sweep issues, filed on espg's direction; #73, #88, and #97 are closed in favor of this issue. Every item from all three was re-verified against current main (post PR #106 merge). Resolved items are listed once for the record; everything still open is grouped by what it needs next, with a priority tag per item.

Resolved since the sweeps (no action — for the record)

• on_minor_arc FP sensitivity → implemented in PR SoS-harden on_minor_arc in robust_crossing #87, then superseded by the uniform predicate in PR Uniform symbolic crossing predicate: fix base-cell-boundary mis-fill #106; on_minor_arc FP/SoS hardening #78 carries a close recommendation.
• CI versioning automation → PRs Tag-driven Cargo.toml version sync #91/Harden release changelog/version commit against non-tip tags #95 (tag-driven changelog/version carryover; Harden release changelog/version commit against non-tip tags #92 closed).
• mort2norm FORCE_PYTHON remnant + stale FORCE_PYTHON benchmark scripts → removed.
• from_latlon point-encode (points=True) → landed via issue numpy-level point-kind geo2mort: geo2mort(lat, lon, order=29, points=True) #96 (phases 1–5 on main).
• Arrow → pandas __from_arrow__ hook → wired (mortie/morton_index.py).
• Pre-emptive flat-cover size guard → moc_to_order guards up front (issue pre-emptive flat-cover size guard #80 / PR small fixes 2026-06-27: densify size guard (#80) + morton_polygon determinism tests (#83) #85).
• smallvec for Node.relevant → signed off and in (src_rust/src/coverage.rs).
• arro3-core CI leg for the pyarrow-free consumer → arro3-no-pyarrow job added (issue arro3-core CI leg for the pyarrow-free consumer was designed but never wired up #98) and green on every PR since.
• Morton __repr__, split_base_cells ordering + wrapper decisions → resolved per Follow up items for the week of July 4, 2026 #97's re-verification.
• PR Tag-driven Cargo.toml version sync #91 mktemp+mv vs sed -i portability sign-off → moot (shipped deliberately; noting closes it).

Group A — small-fixes, bundleable into ONE claude/small-fixes-<date> PR

(each has a standing recommendation; a 👍 on this issue + small-fix labels activate the bundle)

• [low] from_geometry(..., normalize=False) is silently a no-op for LineString/MultiLineString — raise ValueError in the linear branch, symmetric with how moc/tolerance/max_cells are rejected there (mortie/geometry.py ~262) + test. (from Follow up items for the week of July 4, 2026 #97 item 7)
• [low] Clippy useless_conversion false positives (31 at the Follow up items for the week of July 4, 2026 #97 count, ~38 now) — add #[allow(clippy::useless_conversion)] at the crate root with a one-line comment that the pyo3/numpy ? conversion is load-bearing. Has bounced through #34 §D cleanup + fmt/clippy sweep #57 → Follow up items for the week of June 23, 2026 #73 → Follow up items for the week of June 27, 2026 #88 → Follow up items for the week of July 4, 2026 #97 undecided; this is the recommended (a). (from Follow up items for the week of June 23, 2026 #73 item 8)
• [medium] Dissolve winding-sign ambiguity near/over a hemisphere — add the runtime guard that raises when |Σ areas| approaches 2π (fail loud instead of silently inverting exterior/hole on polar-scale covers) + a near-hemisphere test; the comment-only status quo stays risky. Recommended (a) from Follow up items for the week of July 4, 2026 #97 item 5; the full winding-free classifier remains a possible follow-up if polar dissolves become a real workload.

Group B — needs an espg decision before any code

• [medium] // floor-division Morton coarsening operator — fork unchanged since Follow up items for the week of July 4, 2026 #97: (a) ship __floordiv__ alone now, or (b) design the full NEP-42 operator grammar in one discuss so a partial grammar isn't frozen into 1.x. Standing recommendation: (b), given the 1.x freeze contract. (from Follow up items for the week of June 23, 2026 #73/Follow up items for the week of June 27, 2026 #88/Follow up items for the week of July 4, 2026 #97)
• [medium] Earthdata-gated notebooks not publicly Binder-runnable (3 of 5) — CLAUDE.md §4 compliance gap: (a) cache a small anonymous data subset in-tree, (b) rewrite the three notebooks against public endpoints, or (c) document them as authenticated-only exceptions. Needs your read on which; (a) was floated on PR Update example notebooks for order-29 packed encoding + binder wheel #76 with no decision. (from Follow up items for the week of June 27, 2026 #88 item 4)
• [low] moc_to_order size-guard pair from PR small fixes 2026-06-27: densify size guard (#80) + morton_polygon determinism tests (#83) #85 — (i) keep ValueError (recommended: it's a caller-controlled argument condition, not a failed allocation) vs MemoryError; (ii) keep the hard raise pinned to the 1<<20 warn line vs a separate higher ceiling (recommended: separate, higher). One-pass answers; then it's a trivial diff that can ride the Group-A bundle. (from Follow up items for the week of July 4, 2026 #97 item 8)
• [low] prefix_trie.morton_polygon tie-break non-determinism (equal-efficiency cells since the #34 §D cleanup + fmt/clippy sweep #57 max-heap) — document as implementation-defined, or pin with a determinism test. Never individually tracked; cheap either way once you pick. (from Follow up items for the week of June 23, 2026 #73 item 9)

Group C — in flight, no action here

• [high, awaiting merge] res2display() order cap + km/m/cm ladder → draft PR Fix res2display() order cap and add km/m/cm unit ladder #102 (Closes #99), all phases complete, CI green, waiting on your review since July 5. Merging it discharges the last high-priority sweep item.

Suggested sequencing

1. Merge PR Fix res2display() order cap and add km/m/cm unit ladder #102 (Group C — review-ready).
2. One-pass answers on Group B's two low items (they then ride the bundle).
3. small-fix go-ahead on Group A → one bundled PR (Closes per item).
4. The two medium Group-B forks (// grammar, notebooks) as their own discuss/plan when you want them.

Refs #73, #88, #97 (closed into this issue).

[espg]

Merge PR #102 (Group C — review-ready).

already merged

...still considering group B items.