MarkerClusterer does not work for dataset associated with one specific client on our website about js-markerclusterer HOT 6 OPEN

It might be a long shot, but do you have some markers with the 0, 0 values as the position ? I've got the same issue and made a PR for it #892.

0_0

I filtered out all markers with lat=0 and long=0 and it worked.

from js-markerclusterer.

Just curious - are there any "irregularities" (size or bounds) with the dataset for the specific client?

from js-markerclusterer.

Just curious - are there any "irregularities" (size or bounds) with the data for the customer that the clusterer does not function?

Bounds? No. The number of markers being manipulated is less than other clients. Overall memory size? Unknown. One thing that I do notice is that the map for the problem client loads a lot more slowly than the map for other clients. I will check.

from js-markerclusterer.

I don't know if I am doing it correctly, but the memory allocation for the markers array in the Chrome "Memory" tab Allocation Timeline measurement looks to be about 2.5 times what it is for a client of similar number of markers (app. 60 MB versus app. 26MB)

To confirm I have the right number, how would you suggest I grab the memory size of the markers algorithm?

from js-markerclusterer.

You would need a memory profiler to verify. Sounds like there may be a lot more data associated with the markers for the specific client (just guessing -- would need a deeper dive into the data set).

From what I have noticed since Google maps "upgrade" to AdvancedMarker elements is that the number of markers that can be created (compared with "classic" markers) has gone down dramatically. For example, using "classic" markers, I was able to create 6 million markers on an older Windows 10 machine running an Intel i7 processor with 32GB of RAM. Using the same machine and test with advanced markers, Chrome becomes unresponsive at around the 260,000 markers, which leads me to think that the advanced markers are much more "memory heavy" than the classic markers (but Google would have to confirm this).

Also, there is an inherent flaw in the design of the clusterer because it uses map markers as input - which is a colossal waste of memory - as relatively few "map markers" are displayed in the map viewport relative to the overall (large) dataset. This is illustrated in a test I created regarding Creating data points vs creating Google map markers.

from js-markerclusterer.

It might be a long shot, but do you have some markers with the 0, 0 values as the position ? I've got the same issue and made a PR for it #892.

from js-markerclusterer.