1
##########################
2
Frequently Asked Questions
3
##########################
6
Where does the name "Chaco" come from?
7
======================================
9
It is named after `Chaco Canyon <http://www.nps.gov/chcu/>`_, which had
10
astronomical markings that served as an observatory for Native Americans. The
11
original version of Chaco was built as part of a project for the `Space
12
Telescope Science Institute <http://www.stsci.edu/resources/>`_. This is also
13
the origin of the name "Kiva" for our vector graphics layer that Chaco uses for
17
Why was Chaco named "Chaco2" for a while?
18
=========================================
20
Starting in January of 2006, we refactored and reimplemented much of the core
21
Chaco API. The effort has been named "chaco2", and lives in the
22
:mod:`chaco2` namespace. During that time, the original Chaco package
23
("Chaco Classic") was in maintenance-only mode, but there was still code that
24
needed features from both Chaco Classic and Chaco2. That code has finally been
25
either shelved or refactored, and the latest versions of Chaco (3.0 and up) are
26
back to residing in the :mod:`chaco` namespace. We still have
27
compatibility modules in :mod:`chaco2`, but they just proxy for the
28
real code in :mod:`chaco`.
30
The same applies to the :mod:`enable` and :mod:`enable2`
34
What are the pros and cons of Chaco vs. matplotlib?
35
===================================================
37
This question comes up quite a bit. The bottom line is that the two projects
38
initially set out to do different things, and although each project has grown a
39
lot of overlapping features, the different original charters are reflected in
40
the capabilities and feature sets of the two projects.
43
Here is an `excerpt from a thread about this question
44
<https://mail.enthought.com/pipermail/enthought-dev/2007-May/005363.html>`_ on
45
the enthought-dev mailing list.
47
Gael Varoquaux's response::
49
On Fri, May 11, 2007 at 10:03:21PM +0900, Bill Baxter wrote:
51
> Just curious. What are the pros and cons of chaco vs matplotlib?
53
To me it seem the big pro of chaco is that it is much easier to use in a
54
"programatic way" (I have no clue this means something in English). It is
55
fully traited and rely quite a lot on inversion of control (sorry, I love
56
this concept, so it has become my new buzz-word). You can make very nice
57
object oriented interactive code.
59
Another nice aspect is that it is much faster than MPL.
61
The cons are that it is not as fully featured as MPL, that it does not
62
has an as nice interactively useable functional interface (ie chaco.shell
63
vs pylab) and that it is not as well documented and does not have the
66
I would say that the codebase of chaco is nicer, but than if you are not
67
developping interactive application, it is MPL is currently an option
68
that is lickely to get you where you want to go quicker. Not that I
69
wouldn't like to see chaco building up a bit more and becoming **the** reference.
71
Developers, if you want chaco to pick up momentum, give it a pylab-like
72
interface (as close as you can to pylab) !
78
Peter Wang's response (excerpt)::
80
On May 11, 2007, at 8:03 AM, Bill Baxter wrote:
82
> Just curious. What are the pros and cons of chaco vs matplotlib?
84
You had to go and ask, didn't you? :) There are many more folks here
85
who have used MPL more extensively than myself, so I'll defer the
86
comparisons to them. (Gael, as always, thanks for your comments and
87
feedback!) I can comment, however, on the key goals of Chaco.
89
Chaco is a plotting toolkit targeted towards developers for building
90
interactive visualizations. You hook up pieces to build a plot that
91
is then easy to inspect, interact with, add configuration UIs for
92
(using Traits UI), etc. The layout of plot areas, the multiplicity
93
and types of renderers within those windows, the appearance and
94
locations of axes, etc. are all completely configurable since these
95
are all first-class objects participating in a visual canvas. They
96
can all receive mouse and keyboard events, and it's easy to subclass
97
them (or attach tools to them) to achieve new kinds of behavior.
98
We've tried to make all the plot renderers adhere to a standard
99
interface, so that tools and interactors can easily inspect data and
100
map between screen space and data space. Once these are all hooked
101
up, you can swap out or update the data independently of the plots.
103
One of the downsides we had a for a while was that this rich set of
104
objects required the programmer to put several different classes
105
together just to make a basic plot. To solve this problem, we've
106
assembled some higher-level classes that have the most common
107
behaviors built-in by default, but which can still be easily
108
customized or extended. It's clear to me that this is a good general
109
approach to preserving flexibility while reducing verbosity.
111
At this point, Chaco is definitely capable of handling a large number
112
of different plotting tasks, and a lot of them don't require too much
113
typing or hacking skills. (Folks will probably require more
114
documentation, however, but I'm working on that. :) I linked to the
115
source for all of the screenshots in the gallery to demonstrate that
116
you can do a lot of things with Chaco in a few dozen lines of code.
117
(For instance, the audio spectrogram at the bottom of the gallery is
118
just a little over 100 lines.)
120
Fundamentally, I like the Chaco model of plots as compositions of
121
interactive components. This really helps me think about
122
visualization apps in a modular way, and it "fits my head". (Of
123
course, the fact that I wrote much of it might have something to do
124
with that as well. ;) The goal is to have data-related operations
125
clearly happen in one set of objects, the view layout and
126
configuration happen in another, and the interaction controls fit
127
neatly into a third. IMHO a good toolkit should help me design/
128
architect my application better, and we definitely aspire to make
129
Chaco meet that criterion.
131
Finally, one major perk is that since Chaco is built completely on
132
top of traits and its event-based component model, you can call
133
edit_traits() on any visual component from within your app (or
134
ipython) and get a live GUI that lets you tweak all of its various
135
parameters in realtime. This applies to the axis, grid, renderers,
136
etc. This seems so natural to me that I sometimes forget what an
137
awesome feature it is. :)