Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix(time-series table): display null values in time-series table and sortable #19024

Conversation

graceguo-supercat
Copy link

SUMMARY

Currently Time-series table display null value as number 0. I had a fix #18039 but it wasn't handle sorting correctly. in airbnb we found when null value mixed with number, the sort function didn't work, so i reverted the previous PR.

This PR is still try to fix the null value issue with similar approach, and add extra logic fix sorting: if null value is mixed with number values, all null value will be pushed to the smaller end.

TESTING INSTRUCTIONS

CI and manual test

ADDITIONAL INFORMATION

  • Has associated issue: fix: handle null values in time-series table #18039
  • Required feature flags:
  • Changes UI
  • Includes DB Migration (follow approval process in SIP-59)
    • Migration is atomic, supports rollback & is backwards-compatible
    • Confirm DB migration upgrade and downgrade tested
    • Runtime estimates and downtime expectations provided
  • Introduces new feature or API
  • Removes existing feature or API

@codecov
Copy link

codecov bot commented Mar 4, 2022

Codecov Report

Merging #19024 (4bd809b) into master (50bb86d) will decrease coverage by 0.06%.
The diff coverage is 47.61%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master   #19024      +/-   ##
==========================================
- Coverage   66.56%   66.50%   -0.07%     
==========================================
  Files        1641     1643       +2     
  Lines       63495    63455      -40     
  Branches     6425     6448      +23     
==========================================
- Hits        42265    42200      -65     
- Misses      19550    19585      +35     
+ Partials     1680     1670      -10     
Flag Coverage Δ
javascript 51.23% <47.61%> (-0.14%) ⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files Coverage Δ
...rontend/src/visualizations/TimeTable/TimeTable.jsx 0.00% <0.00%> (ø)
superset-frontend/src/utils/sortNumericValues.ts 100.00% <100.00%> (ø)
...ts/nativeFilters/FilterBar/FilterControls/utils.ts 25.00% <0.00%> (-45.00%) ⬇️
...eFilters/FiltersConfigModal/FiltersConfigModal.tsx 64.67% <0.00%> (-14.73%) ⬇️
...d/components/nativeFilters/FilterCard/ScopeRow.tsx 58.82% <0.00%> (-13.91%) ⬇️
...Filters/FilterBar/FilterControls/FilterControl.tsx 70.96% <0.00%> (-11.18%) ⬇️
...tend/src/filters/components/Time/transformProps.ts 50.00% <0.00%> (-7.15%) ⬇️
...d/src/filters/components/Time/TimeFilterPlugin.tsx 47.05% <0.00%> (-6.28%) ⬇️
superset-frontend/src/components/Select/utils.ts 52.94% <0.00%> (-6.15%) ⬇️
...oard/components/nativeFilters/FilterCard/Styles.ts 80.00% <0.00%> (-4.62%) ⬇️
... and 75 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 50bb86d...4bd809b. Read the comment docs.

@graceguo-supercat graceguo-supercat changed the title fix (time-series table): display null values in time-series table and sortable fix(time-series table): display null values in time-series table and sortable Mar 4, 2022
Copy link
Member

@etr2460 etr2460 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

before we make this fix could we either:

  • add unit tests for this function
  • migrate the component to TS

or even both? since we know related changes have broken this chart before. thanks!

@@ -129,7 +129,16 @@ const TimeTable = ({
sortType: (rowA, rowB, columnId) => {
const rowAVal = rowA.values[columnId].props['data-value'];
const rowBVal = rowB.values[columnId].props['data-value'];
return rowAVal - rowBVal;
if (typeof rowAVal === 'number' && typeof rowBVal === 'number') {
Copy link
Member

@ktmud ktmud Mar 4, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This could be simplified as:

return (
  (rowAVal == null) - (rowBVal == null) ||
  Number.isNaN(rowAVal) - Number.isNaN(rowBVal) ||
  Number(rowAVal) - Number(rowBVal)
);

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@ktmud thank you for the simplified expression, but i feel it is a little hard to digest. I would prefer straightforward expression, given there is no significant space or complexity optimization. And after user research, we might want to add extra logic to always push null values at the bottom (instead of sorting as smallest), i think my original expression is easier to update and understood.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

My proposed expression already included logics to sort null and NaN to the bottom.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Push null and NaN to the bottom is not decided yet. I think we will run user research to decide sort or push.

@pull-request-size pull-request-size bot added size/L and removed size/S labels Mar 7, 2022
@graceguo-supercat
Copy link
Author

graceguo-supercat commented Mar 7, 2022

before we make this fix could we either:

  • add unit tests for this function
  • migrate the component to TS

or even both? since we know related changes have broken this chart before. thanks!

i added unit test.
I tried migration, but triggered a lot of errors, I am afraid of causing other bugs. I don't want to lose the focus of this Sorting issue.

Comment on lines 56 to 60
// @ts-ignore
expect(sortFn(rowA, rowB, columnId)).toBe(-1);
// @ts-ignore
expect(sortFn(rowA, rowC, columnId)).toBe(-1);
// @ts-ignore
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why do we need to ts ignore here? if the rows are the improper types, then you can define them as the correct ones above instead of ignoring types

Copy link
Author

@graceguo-supercat graceguo-supercat Mar 8, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Row type is very complicated object, i am not sure how to mock my input data:

export interface UseTableRowProps<D extends object> {
    cells: Array<Cell<D>>;
    allCells: Array<Cell<D>>;
    values: Record<IdType<D>, CellValue>;
    getRowProps: (propGetter?: RowPropGetter<D>) => TableRowProps;
    index: number;
    original: D;
    id: string;
    subRows: Array<Row<D>>;
}

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

const rowA = {} as unknown as UseTableRowProps;?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

wow this trick works!!

if (typeof rowBVal === 'number') {
return -1;
}
return 0;
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

wanna add the test case where both are not a number?

also, can these ever be strings or objects or other types? or only ever number or null?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

return (
  (rowAVal == null) - (rowBVal == null) ||
  Number.isNaN(Number(rowAVal)) - Number.isNaN(Number(rowBVal)) ||
  Number(rowAVal) - Number(rowBVal)
);

This should handle all cases.

Copy link
Author

@graceguo-supercat graceguo-supercat Mar 8, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Given the time-series table viz type, i think the cell values could be string, number, null, number mixed with null.
I agree, this should handle all cases. But Typescript doesn't like

(rowAVal == null) - (rowBVal == null)

any suggestion?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I double checked, the first column metric doesn't use this sorting function. So the cell values could be null, number, no string type.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can wrap them with Number as well. It should convert false to 0 and true to 1.

return (
  Number(rowAVal == null) - Number(rowBVal == null) ||
  Number.isNaN(Number(rowAVal)) - Number.isNaN(Number(rowBVal)) ||
  Number(rowAVal) - Number(rowBVal)
);

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sorry...typescript doesn't like

Number.isNaN(Number(rowAVal)) - Number.isNaN(Number(rowBVal))


const sortNumberWithMixedTypes = (rowA: Row, rowB: Row, columnId: string) => {
const rowAVal = rowA.values[columnId].props['data-value'];
const rowBVal = rowB.values[columnId].props['data-value'];
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we move Row and columnId out of this utility function to make it more generic? I'd imagine it can be useful in other places as well.

import { JSONPrimitive } from '@superset-ui/core';

export default function sortNumberWithMixedTypes(
  valueA: JSONPrimitive | undefined,
  valueB: JSONPrimitive | undefined
) {
}

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sortType is prop from react-table useSortBy: https://github.com/TanStack/react-table/blob/alpha/docs/api/useSortBy.md
I am not sure I can change my function signature?

Copy link
Member

@ktmud ktmud Mar 8, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can always have wrapper function for sortType:

sortType: useCallback((rowA, rowB, columnId) =>
  sortNumberWithMixedTypes(
    rowA.values[columnId].props['data-value'],
    rowB.values[columnId].props['data-value'],
  ), []);

Copy link
Author

@graceguo-supercat graceguo-supercat Mar 8, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ESLint: React Hook "useCallback" cannot be called inside a callback. React Hooks must be called in a React function component or a custom React Hook function.(react-hooks/rules-of-hooks)

this sortType must be memoized.

Copy link
Member

@ktmud ktmud Mar 8, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I know it needs to be memorized, which is why I added useCallback.

You can either move the useCallback up a level or move the wrapper outside of the component as a simple one liner in the component file without using useCallback.

Point is, such utility functions should not depend on too specific input data types if we want to maximize reusability.

@ktmud ktmud force-pushed the gg-fix-time-table-null-values branch from 9475984 to 4bd809b Compare March 9, 2022 00:08
{ descending, nanTreatment: 'asSmallest' },
) *
// react-table sort function always expects -1 for smaller number
(descending ? -1 : 1);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@graceguo-supercat @etr2460 I took a different approach and added the option to change how null/nans are treated. Please take a look

@graceguo-supercat
Copy link
Author

i am ok with new solution, it is a more general comparator sort string/number/NaN/null like a number type way.

Copy link
Member

@etr2460 etr2460 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm, thanks for the tests, the iteration, and all the work here to make sure we don't introduce any more bugs!

@graceguo-supercat graceguo-supercat merged commit d539fc2 into apache:master Mar 9, 2022
villebro pushed a commit that referenced this pull request Apr 3, 2022
…sortable (#19024)

* fix: display null values in time-series table and sortable

* add unit test

* fix unit test

* Add sortNumericValues with different nan treatment

Co-authored-by: Jesse Yang <jesse.yang@airbnb.com>
(cherry picked from commit d539fc2)
@mistercrunch mistercrunch added 🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels 🚢 2.0.0 labels Mar 13, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels lts-v1 size/L 🍒 1.5.0 🍒 1.5.1 🍒 1.5.2 🍒 1.5.3 🚢 2.0.0
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants