EnxDev commented on PR #41490:
URL: https://github.com/apache/superset/pull/41490#issuecomment-4851444746

   ## EnxDev's Review Agent โ€” apache/superset#41490 ยท HEAD 3b96e90
   **lgtm** โ€” clean, test-first bug fix; color logic extracted into a shared 
util and reused unchanged. Two unmentioned standalone-chart behavior changes 
worth a note, both low risk.
   
   This is a refactor + fix: the color-resolution branches move verbatim from 
`CategoricalDeckGLContainer`'s `addColor` into `addColorToFeatures`, and the 
same util is then applied to color-aware sublayers 
(`deck_scatter`/`deck_path`/`deck_arc`) in `Multi.tsx`. I diffed the extracted 
branches against the originals โ€” fixed/categorical/breakpoint cases are 
byte-equivalent. Two deltas:
   
   ### ๐Ÿ”ด Functional
   - **`utils/addColor.ts:111`** ยท _Low_ โ€” the `default` case changed from 
`return []` to `return data`. In the standalone container, `addColor` is called 
with `selectedColorScheme = fd.color_scheme_type` 
(`CategoricalDeckGLContainer.tsx:247`); when that's unset/unrecognized, the old 
code dropped every feature (blank map), the new code returns features untouched 
so the layer renders points in its fallback color. The Multi path can't hit 
this (the call site guards on `color_scheme_type` being truthy), so the change 
is container-only and arguably an improvement โ€” but it's a real behavior shift 
for any saved scatter with no `color_scheme_type`. Confirm `color_scheme_type` 
always defaults for saved charts; if it can be blank, this is observable. 
**regression test:** call `addColorToFeatures(data, fd)` with no/unknown 
`color_scheme_type` and assert the intended outcome (the existing "unrecognized 
scheme returns features unchanged" test pins the new behavior but not the 
container's ol
 d empty-array contract).
   
   ### ๐ŸŸก Should-fix
   - **`CategoricalDeckGLContainer.tsx:73`** โ€” `getCategories` now keys the 
legend on `fd.slice_id` instead of `fd.sliceId`. This is the right call (it 
aligns the legend with the point-color path, which always used `slice_id`, and 
is the canonical `form_data` key โ€” already settled with reviewers). But it also 
changes legend swatch colors for existing standalone categorical scatter charts 
whose scale was previously seeded with `undefined`. It's an unmentioned 
standalone change with no direct test. Worth a line in the PR description and a 
test asserting legend and point colors resolve from the same `slice_id`.
   
   ### ๐Ÿ™Œ Praise
   - `utils/addColor.ts` โ€” extracting the branches into one tested util (and 
the `Multi.color.test.tsx` red-on-master pin) is the right shape; the fix is 
correctly scoped to color-aware layers and leaves others untouched.
   
   Re the `color_picker` alpha drop in the categorical branch (raised by a 
bot): agreed it's lifted verbatim from master and out of scope here.
   
   <!-- enxdev-review-agent:3b96e90 -->
   _Reviewed by EnxDev's Review Agent โ€” @rusackas ยท HEAD 3b96e90._
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to