osm basemap: split the green bucket, texture the ground cover, region as config - #121
Merged
Merged
Conversation
…ke a region a config value Three things, all found from the live deploy. 1. THE FLAT GREEN WASH. `class_for_tags` collapsed landuse|leisure|natural into one `Green` bucket, so `landuse=residential` — which covers most of a suburb — rendered as vegetation. A whole z16 viewport read as meadow. Split into `Meadow` (grass/farmland/heath), `Park` (leisure/cemetery) and `Built` (residential/industrial/commercial, grey and NOT green), with `Green` kept as the fallback so widening the specific set can never demote something to `Other`. Wire codes are APPENDED (7/8/9), never inserted: 0-6 stay a contract with the client's CLASS_ORDER, so an older client renders a new class as unknown rather than as the wrong one. One pre-existing assertion moved: `leisure=park` asserted `Green` because that WAS the collapsed behaviour. It is re-pinned to `Park` with a comment saying the behaviour changed deliberately — not a test edited for convenience. 2. MEADOWS HAD NO TEXTURE (operator). A flat fill cannot tell a meadow from a park from a lawn, and at z16 those are most of the viewport. Canvas gives this cheaply where SVG would not have: one 8x8 offscreen tile per texture, built once, handed to createPattern and reused for every polygon of that class — one pattern object per session, not per shape. Stipple for meadow, canopy strokes for wood. The pattern is deliberately NOT transform- corrected: it is created in canvas space and we translate() per tile, so the texture rides with the map instead of crawling under it while panning. 3. A REGION IS NOW CONFIG, NOT CODE. `ARTIFACTS` was hardcoded to berlin.* and the prefix to q2/bakes/berlin-v1, so a second bake needed a code change. `OSM_BAKE_REGION` now drives artifact names, S3 prefix and the volume path. The name is validated to [a-z0-9-]: it is interpolated into an S3 key AND joined onto a filesystem path, so `..` or `/` would traverse — an invalid value warns and falls back rather than building that path. Because the cache dir is region-keyed, two regions coexist on one volume and flipping between them costs no re-download. Also: the books header's `slab` digest — the same cross-bake pin `.chains` already enforces — was being DISCARDED (`Ok((_header, books))`). A codebook from a different bake was accepted silently, resolving every ordinal to some other bake's string. Now verified, and a magic mismatch is reported as an OUTDATED bake format naming expected vs found, instead of a generic "unreadable" after which every shape silently falls to `ShapeClass::Other`. Verified in headless Chromium against the shipped page, bake endpoints stubbed with the new classes: 2,698 distinct painted colours (was a flat wash), textures visible, 0 SVG nodes, 0 external requests, pan repaints, and the binary wire still 4.7x smaller than JSON on the same shapes. Docs: claude-notes/instructions/osm-bake-new-region.md — the BW runbook with measured sizing (645 MB PBF vs Berlin's 99 MB; ~9.1 GB output; budget 32 GB RAM, which is why BW cannot be baked in the dev container). 103 passed, 0 failed. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NMeiLmtDKhomJNSo2ecbJw
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: serverGenReqId_f1070966-f1c6-45f6-8678-fac60959a3c1) |
|
Caution Review failedThe pull request is closed. ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (4)
📝 WalkthroughWalkthroughThe PR adds operational instructions for baking Baden-Württemberg OSM data, supports region-specific slab hydration, introduces Meadow, Park, and Built feature classes, strengthens sidecar validation, and renders new classes with textured basemap styles. ChangesOSM region and basemap updates
Estimated code review effort: 3 (Moderate) | ~25 minutes Possibly related PRs
Suggested reviewers: Poem
✨ Finishing Touches📝 Generate docstrings
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
AdaWorldAPI
marked this pull request as ready for review
August 12, 2026 09:51
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
2 participants
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.
Three findings off the live deploy. None of these is the bake-format work (see the note at the bottom) — this is the render side plus the config that a second region needs.
1. The flat green wash was my class table
class_for_tagscollapsedlanduse|leisure|naturalinto oneGreenbucket, solanduse=residential— which covers most of a suburb — rendered as vegetation. That is why a whole z16 Zehlendorf viewport read as meadow.Split into
Meadow(grass/farmland/heath),Park(leisure/cemetery),Built(residential/industrial/commercial — grey, not green), withGreenkept as the fallback so widening the specific set can never demote something toOther.Wire codes are appended (7/8/9), never inserted: 0-6 stay a contract with the client's
CLASS_ORDER, so an older client renders a new class as unknown rather than as the wrong one.One pre-existing assertion moved:
leisure=parkassertedGreenbecause that was the collapsed behaviour. Re-pinned toParkwith a comment saying the behaviour changed deliberately — flagging it rather than quietly editing a test that got in the way.2. Meadows had no texture
A flat fill cannot tell a meadow from a park from a lawn, and at z16 those are most of the viewport. Canvas gives this cheaply where the old SVG layer would not have: one 8×8 offscreen tile per texture, built once, handed to
createPatternand reused for every polygon of that class — one pattern object per session, not per shape. Stipple for meadow, canopy strokes for wood.The pattern is deliberately not transform-corrected: it is created in canvas space and we
translate()per tile, so the texture rides with the map instead of crawling underneath it while panning.3. A region is now config, not code
ARTIFACTSwas hardcoded toberlin.*and the prefix toq2/bakes/berlin-v1, so a second bake required a code change.OSM_BAKE_REGIONnow drives artifact names, S3 prefix and volume path.The name is validated to
[a-z0-9-]— it is interpolated into an S3 key and joined onto a filesystem path, so..or/would traverse; an invalid value warns and falls back rather than building that path. Because the cache dir is region-keyed, two regions coexist on one volume and flipping between them costs no re-download.Also: a discarded cross-bake pin
open_booksdidOk((_header, books))— the header'sslabdigest, the same cross-bake pin.chainsalready enforces, was thrown away. A codebook from a different bake was accepted silently, resolving every ordinal to some other bake's string. Now verified. And a magic mismatch is reported as an outdated bake format naming expected vs found, instead of a generic "unreadable" after which every shape silently falls toShapeClass::Other.Verified
Headless Chromium against the shipped page, bake endpoints stubbed with the new classes:
103 tests pass, 0 failed.
Docs:
claude-notes/instructions/osm-bake-new-region.md— the Baden-Württemberg runbook with measured sizing (645 MB PBF against Berlin's 99 MB; ~9.1 GB output; budget 32 GB RAM, which is why BW cannot be baked in the dev container).What this does NOT do
It does not produce a bake in the new format.
EdgeBlockis still written zeroed by the baker (build_row: "edges — zeroed. Reserved, never shrunk."), there are no junction/street-node rows, andBOOKS = 3is pinned so a label codebook is a fourth book and anOSMCBK\0\x02→\0\x03version bump. That work is the actual re-bake and is tracked separately.🤖 Generated with Claude Code
https://claude.ai/code/session_01NMeiLmtDKhomJNSo2ecbJw
Generated by Claude Code
Summary by CodeRabbit