[SciPy-dev] ANN: IPython 0.2.6

[Fernando Pérez]

Hi all,

since there seems to be some consensus that using IPython along with SciPy may
be a good idea, I've made an 'official' release at the usual place with all
recent changes and bugfixes: http://www-hep.colorado.edu/~fperez/ipython/

After discussing things with Eric, we've agreed that right now it's just not
sane to merge fully IPython with SciPy, since both code bases are large,
complex and still in need of work. IPython is in need of a deep internal
rewrite, which I plan to do later this year, hopefully at the end of the
fall. The code is however fairly stable, documented and perfectly useable (I
think) for end users.

I added a 'scipy' profile which does a 'from scipy import *' and has the
beginnings of what we discussed earlier about array output formats. To invoke
it simply type 'ipython -p scipy' (or define a convenient 'scipy' alias to do
it). Doing this makes scipy look like an environment of its own, with all its
functions ready at the prompt. Here's a screenshot :)

[~]> ipython -p scipy
Python 2.2 (#1, Jan 17 2002, 21:03:58)
Type "copyright", "credits" or "license" for more information.

IPython 0.2.6 -- An enhanced Interactive Python.
?       -> Introduction to IPython's features.
?object -> Details about 'object'; object? also works, ?? prints more.
help    -> Python's own help system.
@magic  -> Information about IPython's 'magic' @ functions.
IPython profile: scipy

Welcome to the SciPy Scientific Computing Environment.

In [1]:

The fancy array printing system Eric wanted is still to be implemented, but
all the IPython-specific work is finished. What remains is regular handling of
Numeric arrays into strings, if anyone is up to the task I'll include their
code in future releases :)

All recent bugfixes for Emacs work are in (Travis, please let me know of any
problems) plus other minor changes.

So IPython will probably live a life of its own for a while, but it would be
great if some of you tested it a bit. We might want to make a link in the
SciPy pages to it as a 'companion' package in order to get a wider base of
testers, but I'll leave that decision up to you folks.

Even though the current codebase doesn't really allow for major changes, I've
started a document outlining what needs to be done for the rewrite so all
comments/ideas/suggestions are welcome. This document will be the starting
point for the rewrite, so feel free to add anything you think would make your
'ideal' scientific computing environment (best of
Mathematica/IDL/Matlab/whatever). We'll then see how far we get :)

*Small* features and of course bugfixes will still go in for this series.

One final thing some of you may not know about the system but may find quite
useful: it was designed to be trivial to embed in another python
program. There's an example of how to do it in the docs, basically you create
an instance and then anywhere in your code say

IPythonShell()  # this is your instance

and IPython pops up in your local namespace. This is very useful for debugging
sessions when print isn't enough, or for data analysis where you want to stop,
look at data, start up again a batch part, stop again, etc. If you're familiar
with IDL's stop/.continue pair of commands, that's what this gives you.

Ok, this is long enough. Enjoy, and thanks for the interest.

Best regards,

Fernando

... who crosses fingers waiting for the last minute brown paper bag bug to
blow up :)