search

SINTEF-9012 / PruneCluster

Fast and realtime marker clustering for Leaflet
MIT License
551 stars 133 forks source link

PruneCluster and Meteor.js #101

Closed n0ne closed 8 years ago

n0ne commented 8 years ago

Hi.

I found your package on Atmosphere.js. Do you plan some improvements? I mean server-side rendering or smth like server-side aggregate? What, in your opinion, the best way to send 100,000 markers from Meteor to the client?

fungiboletus commented 8 years ago

Hei,

I am not a meteor developer myself, the package has been made by @mordka.

However yes I have some plans to improve the library, especially on client side API. For server-side rendering/aggregation, I have some long terms ideas about using stream processing with big data toys such as Storm or Spark Streaming but it's unclear.

I don't know how Meteor transfers the data (compressed json?) but I successfully used ProtocolBuffer and ProtoBuf.js in my applications. 100 000 markers is not much when serialized in ProtocolBuffer, but if you have a lot of metadata in strings you may want to compress the protocol buffer as well (it might be automatic if you use websockets for example). Also if you use websockets, I recommend using chunks instead of sending each marker separately.

n0ne commented 8 years ago

Here Meteor example with 32951 markers. It is not very fast )-: And I have no idea, how send chunks in pub/sub schema in Meteor...