🐛 fix duplicate indicators in indicator-based explorers

[ikesau]

A naive potential fix for #3943

I don't understand the Explorer data model well enough to know why explorerProgram.grapherConfig has the yVariableIdsthat it does - but it seems this line of code that inserts extra variables into the grapher config to get it to load the indicator data and metadata was assuming there'd be no duplicates.

There's almost certainly a better upstream fix for this, but this can fix the issue in the meantime until someone who understands explorers better can work things out.

[owidbot]

Quick links (staging server):

Site	Admin	Wizard

Login: ssh owid@staging-site-fix-duplicate-indicators-explorers

• Site-screenshots: https://github.com/owid/site-screenshots/compare/fix-duplicate-indicators-explorers
• SVG tester: https://github.com/owid/owid-grapher-svgs/compare/fix-duplicate-indicators-explorers

SVG tester:

Number of differences (default views): 0 ✅
Number of differences (all views): 0 ✅

Edited: 2024-09-09 19:34:55 UTC
Execution time: 1.16 seconds

[lucasrodes]

Thanks Ike! I was checking and found some views where the color for different indicators was the same. Also, when hovering over one of these, multiple indicators are highlighted.

It looks to me as this doesn't happen when freshly loading a view, but when changing the selection in the drop-down options. I found this behavior in

• Conflict Data Explorer
• Conflict Data Source Data Explorer

Examples

Example 1

• Go to http://staging-site-fix-duplicate-indicators-explorers/explorers/conflict-data
• Change value in first drop-down to have Conflict type = All state-based conflicts
  

Example 2

• Go to http://staging-site-fix-duplicate-indicators-explorers/explorers/conflict-data-source
• Change value in second drop-down to have Conflict type = All state-based conflicts
  

[lucasrodes]

I just compared this site version, with one prior to #3793, and it looks like the coloring issue I explained above was there already. Bastian also mentioned that this coloring issue had happened in the past (slack).

Since there are multiple explorers live affected by this issue (Conflict Data Explorer, Conflict Data Source Data Explorer, Population and demography Explorer, Migration Explorer, COVID Explorer) I suggest we merge this and we can open a separate issue for the color issue. cc @danyx23

[larsyencken]

Suggested change

	!dimensions.some((d) => d.variableId === yVariableId)
	!dimensions.some((d) => d.variableId === yVariableId && d.property === DimensionProperty.y)

Some scatters have the same variable as x and y dimensions, so maybe it's worth being more defensive and only filtering out y duplicates at this spot?

[danyx23]

Looks good! Thanks for the fix, Ike!

[danyx23]

I'll merge this since it leads to errors on today's data insight follow-up link