Plugin
- Plugin id: geochron
- Plugin version: 1.0.3
Both problems are pre-existing on main — confirmed by stashing unrelated changes and re-running, giving identical figures.
1. Overflow at 128×32 (the more serious one)
[FAIL] 128x32 geochron overflow bbox=(128, 1, 130, 21)
The plugin draws 2px past the right edge of a 128-wide panel. That is a hard harness failure rather than drift — content outside the panel is either clipped or, on a chained display, lands on the next panel.
128×32 falls in the wide sidebar layout (aspect ratio 4.0), so the sidebar carrying the UTC/local time and subsolar coordinates is the likely culprit. The overflow is 20px tall starting at y=1, which matches a text block rather than the map.
Notably 256×32 (also wide sidebar, aspect 8.0) does not overflow — so it looks like the sidebar is sized against something that does not scale down correctly at the narrower width.
2. Golden drift on six of eight sizes
[FAIL] 64x32 golden drift: 420px (max Δ=250)
[FAIL] 64x64 golden drift: 420px (max Δ=248)
[FAIL] 128x64 golden drift: 420px (max Δ=245)
[FAIL] 256x32 golden drift: 388px (max Δ=245)
[FAIL] 128x96 golden drift: 420px (max Δ=245)
[FAIL] 256x128 golden drift: 420px (max Δ=245)
[PASS] 96x48
The consistent 420px / Δ≈245 across sizes suggests one element moved or changed colour rather than the map rasterisation shifting — a map change would scale with panel area, and 96×48 would not be the lone survivor.
Why I have not regenerated the goldens
Same reasoning as #371: I cannot tell from here whether the goldens are stale or are correctly catching a regression, and regenerating from my environment risks baking in local font/image behaviour. ChuckBuilds/LEDMatrix#364 already showed one case where my environment differed from yours in a way my checks did not catch.
The overflow is worth fixing regardless of what the goldens say — it fails on its own terms, not by comparison.
Note
The plugin's own render_preview.py is well built — it imports geochron_renderer, the same module manager.py draws with, so previews cannot drift from the plugin. Running it at 128×32 may make the sidebar overflow easy to see.
Found while writing the README in #375.
Plugin
Both problems are pre-existing on
main— confirmed by stashing unrelated changes and re-running, giving identical figures.1. Overflow at 128×32 (the more serious one)
The plugin draws 2px past the right edge of a 128-wide panel. That is a hard harness failure rather than drift — content outside the panel is either clipped or, on a chained display, lands on the next panel.
128×32 falls in the wide sidebar layout (aspect ratio 4.0), so the sidebar carrying the UTC/local time and subsolar coordinates is the likely culprit. The overflow is 20px tall starting at y=1, which matches a text block rather than the map.
Notably 256×32 (also wide sidebar, aspect 8.0) does not overflow — so it looks like the sidebar is sized against something that does not scale down correctly at the narrower width.
2. Golden drift on six of eight sizes
The consistent 420px / Δ≈245 across sizes suggests one element moved or changed colour rather than the map rasterisation shifting — a map change would scale with panel area, and 96×48 would not be the lone survivor.
Why I have not regenerated the goldens
Same reasoning as #371: I cannot tell from here whether the goldens are stale or are correctly catching a regression, and regenerating from my environment risks baking in local font/image behaviour. ChuckBuilds/LEDMatrix#364 already showed one case where my environment differed from yours in a way my checks did not catch.
The overflow is worth fixing regardless of what the goldens say — it fails on its own terms, not by comparison.
Note
The plugin's own
render_preview.pyis well built — it importsgeochron_renderer, the same modulemanager.pydraws with, so previews cannot drift from the plugin. Running it at 128×32 may make the sidebar overflow easy to see.Found while writing the README in #375.