Beaker Notebook has much cleaner support for integrating d3 (or any JS lib) with Python or other languages: <a href="http://sharing.beakernotebook.com/gist/anonymous/e21582541d7c1fe60eb4" rel="nofollow">http://sharing.beakernotebook.com/gist/anonymous/e21582541d7...</a><p>learn more at <a href="http://BeakerNotebook.com/" rel="nofollow">http://BeakerNotebook.com/</a>
The Jupyter web page is a fine example of reverse information distribution. I went from" I think I know what this is" to "I have no idea what this is" within seconds of the page loading.
FYI, the latest Jupyter + matplotlib have interactive plotting inline with "%matplotlib notebook". The server must be running for this to work (so you can't interact with static ipynb files in nbviewer, for example), but a static version of the plot is generated for static notebooks. There are also some changes coming down the line to expose the traits of matplotlib plots so that libraries like mpld3 can be integrated more seamlessly.<p>(Also, Hi Christian!)
Interestingly, Zeppelin (a competitor to IPython/Jupyter mostly aimed at use with Spark) uses D3 for most of its visualizations.<p>IMHO Jupyter is actually easier to get working (even with Spark support) though.<p>[1] <a href="https://zeppelin.incubator.apache.org/" rel="nofollow">https://zeppelin.incubator.apache.org/</a>
IPython feels like an important new way to do computing, particularly for exploring and teaching. I love the idea of bringing more powerful HTML tools into it.