Skip to content

test: re-vendor osm-testdata grid at 98a9b32, drop redundant unit test - #781

Merged
brawer merged 1 commit into
mainfrom
revendor-osm-testdata-769
Aug 29, 2026
Merged

brawer merged 1 commit into
mainfrom
revendor-osm-testdata-769

Conversation

@brawer

@brawer brawer commented Aug 29, 2026

Copy link
Copy Markdown
Owner

Context

When we vendored the osm-testdata grid fixtures (#529, #535), a delta analysis against nimix/osm_conv_tests turned up exactly one geometry case the grid set lacked: a multipolygon whose inner ring touches the outer ring at two nodes, splitting the outer into two disjoint components. We covered that two ways — a standalone unit test in polygon_assembler.rs (#643), and a PR to upstream osm-testdata adding the case as a real fixture.

That upstream PR has landed. osmcode/osm-testdata master (98a9b32) now carries grid test 769 ("Multipolygon with one outer and one inner ring, inner touching outer in two nodes, splitting the outer into two components"), with a valid default MULTIPOLYGON oracle.

Changes

  • Re-vendor tests/test_data/osm-testdata-grid/ at 98a9b32aafe8798523d8b3819c33199eee6d5687. The diff from the previously-vendored ecfdeb0d is only fixture 7/769/ plus the regenerated all.osm / tests.json aggregates — nothing else upstream changed.
  • Delete polygon_assembler::tests::inner_touching_outer_at_two_points_splits_it_into_two_pieces. It existed only to fill the gap that 7/769 now fills; the grid harness exercises the case against 7/769's oracle through the production GeometryBuilder path.
  • The other polygon_assembler tests that mirror grid fixtures (7/707/708/733/749, 7/790, 7/792, 7/795) stay — those were written as deliberate unit-level checks of PolygonAssembler's API, not gap-fillers.
  • SBOM needs no change: scripts/sbom/generate-sbom.sh reads the commit from VENDORED.md.

Verification

  • geometry::grid_tests::grid_multipolygon_tests passes with 7/769 now exercised; no_oracle_fixtures_are_still_unverifiable passes (769 correctly stays out of NO_ORACLE_FIXTURES).
  • Full cargo test green (lib, integration, doctests).
  • cargo clippy — only pre-existing warnings in src/pipeline/osm/assemble.rs, untouched here.

🤖 Generated with Claude Code

Upstream osm-testdata now carries grid fixture 7/769 ("Multipolygon with
one outer and one inner ring, inner touching outer in two nodes,
splitting the outer into two components"), the one geometry case our
delta analysis against nimix/osm_conv_tests found the grid set lacked
(added in #643 as a standalone unit test, and submitted upstream at the
same time).

Re-vendor grid/data at osmcode/osm-testdata master
(98a9b32aafe8798523d8b3819c33199eee6d5687) to pick up 7/769 -- the only
fixture change since the previously-vendored ecfdeb0d -- and delete
polygon_assembler's inner_touching_outer_at_two_points_splits_it_into_two_pieces,
which existed only to cover that gap. The grid harness now exercises the
case against 7/769's default MULTIPOLYGON oracle via the production
GeometryBuilder path.

The other polygon_assembler tests that mirror grid fixtures stay: they
were written as deliberate unit-level checks of PolygonAssembler's API,
not as gap-fillers.

SBOM needs no change -- generate-sbom.sh reads the commit from VENDORED.md.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MqAiLYcfXhVmDGxXgwYHGj
@brawer
brawer enabled auto-merge August 29, 2026 06:55
@brawer
brawer added this pull request to the merge queue Aug 29, 2026
Merged via the queue into main with commit 53393d4 Aug 29, 2026
6 checks passed
@brawer
brawer deleted the revendor-osm-testdata-769 branch August 29, 2026 07:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant