Closed lmorroni closed 10 years ago
I agree one should be able to pass any geojson
through javascript to anycluster. We have to keep in mind that anycluster expands the given viewport to the closes possible set of squares of the world grid. I think this is not necessary for a drawn polygon because with you want the contents of this exact polygon to be clustered. As the polygon does not change when panned or zoomed the clustering is easier. If we pass geojson
to getCluster
, which I think is a good idea, we would have to determine somehow if a world grid should be used or not.
My first idea would be to pass an optional parameter to getCluster
alongside the geojson
which enforces the use of the exact geometry without snapping it to the world grid (this parameter would also enforce consistency between cluster count/ pins-on-viewport-count at the cost of a fixed grid).
To make the amount of possible clusters consistent, the gridsize
should be mathematically compared with the size of the area covered by the polygon
to determine the k
of the kmeans algorithm
.
If you have any further thoughts let me know.
I am rewriting anymals.org since a few months and will get to the maps soon, so I will have enough time to implement this feature.
viewport now is geojson
Hi, Wouldn't it be better if the
getCluster
javascript function took ageojson
representation of theviewport
? It already does a conversion before calling theclusterMethod
view. This would add the ability to for us to pass in a drawn polygon. I need this functionality for users who want to see clusters/placemarks for a specific area such as along a river. I am adding the Google Maps drawing library so a user can trace out a specific area of the map and only get clusters from that area. What do you think? Larry