fix(core): let GlobeController zoom past level 12

[dchaudhari7177]

Closes #10736

Background

Above zoom 12, GlobeView renders with a WebMercatorViewport. That class has no getZoomAnchorStrength, so a pointer-anchored wheel zoom threw in GlobeController.applyConstraints, and the view couldn't zoom in any further. Now only a GlobeViewport is asked for its anchor strength. A WebMercatorViewport's map covers the whole screen, so the anchor holds at full strength there.

Change List

• GlobeController.applyConstraints: check instanceof GlobeViewport before calling getZoomAnchorStrength
• Unit tests: a wheel zoom from 12.5 zooms in and keeps the pointer's coordinate (to 4 decimals). Without the fix it fails with viewport.getZoomAnchorStrength is not a function.

Checked locally: controllers.spec.ts (86 tests) passes, tsc --noEmit -p modules/core passes, and prettier is clean. The spec normally runs in the headless project, but I don't have Playwright's Chromium on this machine, so I ran it through the node project from a temporary *.node.spec.ts copy. CI will run it in headless.

---

Note

Low Risk
Small, targeted change to globe zoom anchor logic with new unit tests; no auth, data, or broad API surface impact.

Overview
Fixes GlobeController pointer-anchored wheel zoom above zoom level 12, where GlobeView switches from GlobeViewport to WebMercatorViewport (#10736).

getZoomPosition no longer assumes a globe viewport or calls getZoomAnchorStrength on every viewport. For a WebMercatorViewport it keeps zooming around the pointer (full-screen map = full anchor strength); globe-only off-globe fallback behavior is unchanged.

Adds controller tests for wheel zoom starting above 12.5 and for a single gesture crossing the zoom-12 handoff, asserting zoom increases and the coordinate under the pointer stays stable.

^{Reviewed by Cursor Bugbot for commit 9a83120. Bugbot is set up for automated code reviews on this repo. Configure here.}

[greptile-apps]

Confidence Score: 5/5

[Medium impact] The production fix appears safe to merge; the only remaining concern is non-blocking test precision at the zoom-12 viewport handoff.

Findings

1.  Viewport handoff remains untested ▶

Summary

This PR allows pointer-anchored GlobeController wheel zoom beyond zoom 12, when GlobeView switches to WebMercatorViewport.

• Calls the globe-specific anchor-strength method only for GlobeViewport instances.
• Preserves full-strength pointer anchoring for WebMercatorViewport.
• Adds regression coverage above zoom 12 and across the viewport handoff.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  A[Pointer wheel zoom] --> B{Resolved viewport}
  B -->|GlobeViewport| C[Check globe anchor strength]
  C -->|On globe| D[Zoom around pointer]
  C -->|Off globe| E[Fall back to center]
  B -->|WebMercatorViewport| D

Loading

_{Reviews (4) · Last reviewed commit: "test(core): cover the GlobeView zoom 12 ..." · Reviewed by Greptile}

[greptile-apps]

Viewport handoff remains untested

This test begins at zoom 12.5 and zooms further in, so it uses WebMercatorViewport throughout. It does not cover the zoom-12 handoff, where the viewport changes and anchor strength switches from the globe-specific calculation to 1. A pointer jump during the first wheel event that crosses this boundary could therefore escape the regression suite. Please add a case that starts at or below zoom 12, crosses above it, and verifies that the pointer coordinate remains stable.

Knowledge Base Used: Distribution and quality

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[dchaudhari7177]

Heads-up: #10713, merged on 09-28, reverted #10598. That revert removed the getZoomAnchorStrength call this PR guarded. GlobeState.applyConstraints on master now does viewport.panByPosition(...), and both GlobeViewport and WebMercatorViewport have that method. So the #10736 crash (n.getZoomAnchorStrength is not a function) should no longer reproduce on master, and this branch now conflicts with code that no longer exists.

I see two options:

1. Close this PR and close [Bug] GlobeView can not zoom in past 12 zoom - n.getZoomAnchorStrength is not a function #10736 as fixed by the revert, or
2. Rescope it to a regression test only. It would keep the wheel-zoom-above-12 case and add the zoom-12 handoff case Greptile asked for (start at 11.9, cross 12, assert the pointer's lng/lat stay put), so a later re-land of fix(core): rotate GlobeView like a ball through poles #10598 can't bring the crash back.

I'm happy to do either. Let me know which you'd prefer. If I don't hear back, I'll close it in a few days rather than leave a stale conflict.

[chrisgervang]

Hey @dchaudhari7177, thanks for working on this and re-evaluating after we reverted #10598. I'm in favor of adding the regression tests you've described and re-purposing this PR for that.

Please let me know when the tests are up and I'll review.

[coveralls]

coverage: 83.718% (+0.02%) from 83.703% — dchaudhari7177:fix/10736-globe-zoom-past-12 into visgl:master