Totally misleading headline. This is using clustering so there are not 100000 pins displayed but low numbers of them. I don't mean to belittle the speed of the clustering algorithms (I have no idea) but I expected something much different.

Well, 100'000 pins displayed simultaneously on a map wouldn't be very readable, would it? Out of curiosity, what did you expect?

Yeah I was initially thinking of something like desktop Google Maps where if you search "food manhattan" you get hundreds of clickable points. Obviously as you said that wouldn't scale (in a user experience sense) to hundreds of thousands though!

Clustering isn't exactly new or exciting. I was expecting what was mentioned in the title, actually.

For another approach to this problem, see the Urbanspoon iPad app. It shows every restaurant in a city, drawing them as dots in a map overlay then swapping for pins when you zoom in enough that a manageable number would be shown.

This doesn't deal with clustering, of course. It's a different UI solution.

Which specific dataset from Paris' Open data are you mapping?

This one: http://opendata.paris.fr/opendata/jsp/site/Portal.jsp?docume...

You should package an app out of that. iTrashcan? :)

Good to see companies sharing great code! Thanks Applidium!

You're welcome! Don't hesitate to submit bug reports or enhancement requests :-)

This is fantastic. Opens up all kinds of new ways to use maps when you can have this much location specific data without cluttering or slowing the UI.

There is also Marker Clusterer for this: http://google-maps-utility-library-v3.googlecode.com/svn/tru...

I very much like this fresh solution though, particularly fond of the animation.

Thank you! To be honest the idea isn't ours: Apple does it already in their Photo apps. But for some reason they never opened this up in the SDK, so we decided to write and publish our own solution.

I'd be curious to see how Apple's code would handle 100'000 photos though :-)