Clustering items

[next2heaven]

Been testing the new clustering. Looking good! The only 2 issues I see is that depending on event heights it's doesn't bucket properly. I think the calculation being used is expecting each row to be a certain height so it ends up hiding some items when you have variable hight items on the timeline.

The other thing is having an option to click the bucket and the timeline will zoom till all items in the bucket can safely be shown. I know that's on the list.

This is such a big item. Glad to see it worked on. Keep up the great work!

[josdejong]

Thanks. This first solution does indeed not take into account the size of the items, but only their timestamp, and does not take into account the stacked height at all. This makes clustering easy and relatively fast: you can filter and cluster before rendering items.

It is an interesting problem: suppose you have 1m items. If you do take into account their actual size, you will have to render them all once in order to determine their size. Only after that you can determine what the stacked height will be, and and cluster them accordingly. But that rendering is exactly the most expensive part.
A solution can be to specify a default size for the items, which more or less corresponds to the actual size, and use this default size as an approximation as long as the item hasn't been rendered.