Moving a marker throws an error and marker disappears

[AsamK]

Moving a marker inside a MarkerCluster.LayerSupport layer with marker.setLatLng() produces an error:

error TypeError: Cannot read property 'x' of undefined
    at L.DistanceGrid._sqDist (https://unpkg.com/[email protected]/dist/leaflet.markercluster-src.js:1917:20)
    at L.DistanceGrid.getNearObject (https://unpkg.com/[email protected]/dist/leaflet.markercluster-src.js:1899:20)
    at NewClass._addLayer (https://unpkg.com/[email protected]/dist/leaflet.markercluster-src.js:984:36)
    at NewClass.<anonymous> (https://unpkg.com/[email protected]/dist/leaflet.markercluster-src.js:253:11)
    at NewClass.addLayers (https://unpkg.com/[email protected]/dist/leaflet.markercluster-src.js:287:4)
    at NewClass.addLayers (https://unpkg.com/[email protected]/dist/leaflet.markercluster.layersupport-src.js:115:8)
    at NewClass._processSingleAddRemoveBuffer (https://unpkg.com/[email protected]/dist/leaflet.markercluster.layersupport-src.js:267:29)

See: https://playground-leaflet.rhcloud.com/cave/edit?html,output

The problem is that LayerSupport buffers the add/remove operations and MarkerCluster implements moving a marker via removing and readding the marker. MarkerCluster expects the remove/add calls to be synchronous:

	_moveChild: function (layer, from, to) {
		layer._latlng = from;
		this.removeLayer(layer);

		layer._latlng = to;
		this.addLayer(layer);
	},

The issue can be worked around by disabling buffering in LayerSupport (by setting singleAddRemoveBufferDuration: 0).
See: https://playground-leaflet.rhcloud.com/yayab/edit?html,output

I'm not sure if this issue can be fixed in LayerSupport, or needs an improved move behaviour in MarkerCluster.

[ghybs]

Hi,

Thank you for the detailed report!

Hum I see the issue…
I cannot think of an easy solution to make _moveChild compatible with singleAddRemoveBufferDuration right now.

As you say, the easy workaround is to set singleAddRemoveBufferDuration option to 0.
Maybe it could be even better to have 0 as the default value?
Or at least mention it in the README?

[AsamK]

Hm, I think making 0 the default value would be good then. It's better to work correctly by default than optimize for performance. The singleAddRemoveBufferDuration option should be documented to have issues with moving markers.

[ghybs]

Shipped in version v2.0.0