General Questions
How do you pronounce “bokeh”?
Either “boh-Kay” or “boh-Kuh”. For more information, you can
consult Wikipedia or an actual
Japanese person.
Why did you start writing a new plotting library?
There are a number of reasons why we wrote a new Python library, but they
all hinge on maximizing flexibility for exploring new design spaces for
achieving our long-term visualization goals. (Please see
Technical Vision for details about those.)
The client-server nature of displaying for the web is reflected in the
fundamental architecture of the library, and also has significant impact on
the front-end interface. Additionally, the difficult task of coalescing
large datasets for display on thin clients in a perceptually useful way
demands different things of a graphics and rendering architecture than most
of the readily-available libraries in Python today.
In the meantime, we would like to remain as compatible with the rest of
the Scipy and PyData ecosystems of tools as possible. This means that we
are very eager to get contributions that help us understanding
configuration files from other libraries, provide API compatibility layers,
and possibly create backends for other libraries on top of Bokeh’s
low-level glyph API.
Please see Contact and get in touch with the dev team if you have
ideas along these lines.
What is the relationship between Bokeh and Chaco?
There is no direct active relationship between these two projects. Some of
the design choices in Chaco are reflected in the architecture of Bokeh, and
some snippets of code from Chaco have been ported to Javascript and placed
in BokehJS. The goals of the two projects are quite different. If you
have a Chaco project that you’d like to put on the web, there is a good
chance that you can make it work with Bokeh (since the HTML5 Canvas API is
fairly close to Kiva’s), but at this point, you will need to write
Javascript if you want custom interactors. For rich client, customizable,
interactive visualization in Python, Chaco is still a good tool.
How does Bokeh compare to mpld3?
For a lightweight, python-only library that exposes most of matplotlib
to the browser, mpld3 could be a good choice. Bokeh also intends
to fully support the MPL interface (and hence Seaborn, pandas, and
ggplot.py), however the main goal of Bokeh is to provide approachable
capability for novel interactive visualizations in the browser. If you
would like to have the benefits of HTML canvas rendering, dynamic
downsampling, abstract rendering, server plot hosting, and the possibility
of interacting from languages besides python, please consider Bokeh for
your project.
Does Bokeh implement R’s ggplot2?
No, Bokeh does not implement R’s ggplot2 interface. Bokeh was influenced
by Wilkinson’s Grammar of Graphics (e.g. the idea of tight, well-defined,
composable abstractions that can be tied directly to data). However, Bokeh
does not implement a “GoG” system. Please see Technical Vision
for more details.
An experimental ggplot interface was prototyped at early stages, (circa
Bokeh 0.1), but it was never fully developed and has since been removed.
At present there are no plans to implement R’s ggplot2 interface directly
within Bokeh. However, since Bokeh provides some level of Matplotlib
compatibility, it is often possible to convert plots created using the
third-party python library ggplot.py
into Bokeh plots very easily. There are several examples in the gallery;
you can see one here.
Technical Questions
Does Bokeh use D3.js?
No. D3 is very cool and its predecessor Protovis was one of the
inspirations for Bokeh. However, we understand the goals of D3 to be about
providing a Javascript-based data scripting layer for the DOM, and this is
somewhat orthogonal (at this point) to the visualization challenges that
Bokeh is trying to tackle. Please see Technical Vision for more
details about the underlying goals and vision behind our project.
Why not use something like Vega?
We may very well end up using the Vega grammar. For now, we need to be
able to specifically tag certain objects with UUIDs so that the object
graph structure can be reconstituted on the JS side. Additionally, we
use this JSON to reproduce Python object graphs when we load up a
Python plot from the plot server - so we would need to make sure that Vega
can fully encapsulate all the information we need here as well.
How do I get the sample data?
Some of the Bokeh examples rely on sample data that is not included
in the Bokeh GitHub repository or released packages, due to their
size. The sample data can be obtained by executing the following
commands at a python prompt:
>>> import bokeh.sampledata
>>> bokeh.sampledata.download()
