Brushing and linking data structures are not updated properly on filtering

[mfranke93]

Setting filters (confidence has filters by default) and then brushing a religion will sometimes link places where those religions do not exist under the applied filters. Specifically, it seems the data structures for storing the relationships between the data attributes are not properly updated on filtering, but should be.

Example (production server data): Default confidence filters apply, also some religion filters, which however do not matter in this case. Brushing SYR links Samarqand (which has SYR without filters, but not under the current filters):

Visualization state:

{"show-filtered":true,"display-mode":"religion","timeline-mode":"qualitative","map-mode":"clustered","confidence-aspect":"religion_confidence","map-state":{"zoom":6,"center":{"lat":38.013476231041935,"lng":67.50000000000001},"base_layer":"light","overlay_layers":["markerLayer"]},"filters":{"religion":{"type":"simple","filter":[5,4]},"time":null,"sources":null,"confidence":{"time_confidence":["certain","probable","contested","uncertain","false",null],"location_confidence":["certain","probable","contested","uncertain","false",null],"place_attribution_confidence":["certain","probable","contested","uncertain","false",null],"source_confidences":["certain","probable","contested","uncertain","false",null],"interpretation_confidence":["certain","probable","contested","uncertain","false",null],"religion_confidence":["certain","probable","contested","uncertain","false",null]},"tags":true,"location":null,"places":null},"metadata":{"version":"v1.0.5","createdBy":"max","createdAt":"2022-03-10T08:30:08.588Z","source":"filesystem","evidenceCount":4564}}

[tutebatti]

will sometimes

I assume we have no idea how often this is the case? This would matter in deciding if this is urgent or not.

[mfranke93]

I assume we have no idea how often this is the case? This would matter in deciding if this is urgent or not.

I suspect this happens most of the time, but is only noticeable in some specific cases where the number of evidences per city/glyph is small enough, such as the case above.

Nevertheless, I have time to take a look at it now, so it should hopefully be fixed in time for the public release.