Split treemaps?

[mbostock]

As described in “Ordered and Unordered Treemap Algorithms and Their Applications on Handheld Devices” by Engdahl.

[timelyportfolio]

inspired by @thomasp85 ggforce implementation and the original paper, I have assembled an ugly mix of code from these two sources and other d3 treemap code https://github.com/timelyportfolio/d3-hierarchy/tree/feature/split-treemap. Before I submit pull with proper tests, I wanted to first make sure that this (with much revision of course) would be welcome.

@mbostock I forked your d3 source treemap (thanks so much for this!) and applied the new split algorithm http://bl.ocks.org/timelyportfolio/e480359ea953d96a38a4466187d3d62b.

In just a little bit, I will also apply the d3v4 treemap with tile input to more closely inspect differences.

[mbostock]

Fantastic! Implementation looks relatively straightforward which is always nice! Thank you for doing this and I look forward to the PR.

[timelyportfolio]

Just converted this to add treemapSplit http://bl.ocks.org/timelyportfolio/2bbd4cedf597eae6a5e90412d9df1c1f. (Had no idea d3.hierarchy had changed so much since this early iteration). I absolutely love all the convenience functions added in d3.hierarchy.

[timelyportfolio]

started and it seems to work correctly on smaller test networks and the d3 source example with much better speed, but now on my test on a larger network I get

Uncaught RangeError: Maximum call stack size exceeded(…)

I'll research and debug (probably completely rewrite).

[timelyportfolio]

Certainly no rush, but is this something you would like me to continue to push forward? Thanks for everything, and please, please, please let me know if ever in any strange little way whatsoever I can help you in any way.

[mbostock]

Turns out—with a bug fix and a small improvement—these are the same thing as binary treemaps which were implemented in d3-hierarchy@1.0 (d3@4.0)!

[timelyportfolio]

👍 good learning experience. Thanks!