You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Bug]: a globe locator just over the smallest globe is clipped near a pole though a centre where it fits exists #2394
I searched the existing issues and did not find a duplicate.
This is a bug report, not a usage question.
What happened?
A Publication Layout's globe locator a little larger than the smallest globe the camera can draw is clipped near a pole, even though a centre exists where the whole globe fits its circle with the area on it. Found in review of #2355.
When the camera is at its coarsest resolution, the disc it draws is about 17.8 mm across on the equator and grows as one over the cosine of the latitude. So an 18 mm locator holds the whole globe at any centre up to about 8.6° of latitude, and from there an area at 83.5° N is about 75° away, in front of the horizon (about 80° from the centre). But the search for the locator's centre (drawable_globe_center, used by globe_locator_center in crates/engine/src/cmd/layout.rs) only accepts latitudes where the camera is under its coarsest resolution. Once no latitude passes, it falls back to the equator, where the area is behind the rim, so the locator keeps the area in the middle with the globe clipped and the "too small to show the whole globe" notice.
This is no worse than before #2355 (the locator was clipped there too, without a notice), and it only affects locators between about 17.8 and 19.4 mm, a small figure's.
Steps to reproduce
Open the World Atlas sample.
Run layout.publication with {"profile": "figure-1col", "extent": [10, 83, 30, 84], "locator": "globe"}.
Look at the locator, and at the layout's warnings.
Expected behavior
The locator is centred as near the area as the whole globe fits its circle, the area on the disc, with the recentring notice instead of the clipping one. The search would test the disc drawn against the circle rather than the camera's limit, and check the horizon at the centre it finds. It keeps the clipped fallback when no centre fits or the nearest one still hides the area. The figure-1col case in a_globe_locator_near_a_pole_fits_its_circle, which now expects clipping for this input, changes with it. drawable_globe_center also places the whole-globe frame, so that path needs checking too.
I’ll take this. I’ll use the rendered disc size to find the closest fitting locator center, retain the horizon check and clipped fallback, and verify the 18 mm publication layout and whole-globe frame regressions.
Activity
choupeanut commentedon Nov 18, 2026
I’ll take this. I’ll use the rendered disc size to find the closest fitting locator center, retain the horizon check and clipped fallback, and verify the 18 mm publication layout and whole-globe frame regressions.