[IPython-dev] using the notebook for teaching?

[Thomas Kluyver]

On 5 December 2012 22:24, Emmanuelle Gouillart <
emmanuelle.gouillart at nsup.org> wrote:

>         My main concern is about stability. As I see it, a professor
> could have a set of different notebook files corresponding to the
> different sessions, and encourage the students to do the same. However,
> when I have tried to open existing .ipynb files, I've run across an
> incompatibility between ipython 0.12 and notebooks in version 3 format
> (error message "Unreadable JSON notebook"). I have not tried whether
> backward compatibility works, ie whether ipython 0.13 can read version 2
> format. Is it planned that the notebook format will still change a lot or
> not? Indeed, I do not expect the undergraduate schools to update ipython
> very frequently, and it would be quite a problem if a professor cannot
> grade a student's notebook because it was written on a more recent
> Ipython... On the other hand, the notebook is really cool and has a lot
> of advantages for teaching, so I really have mixed feelings... Stability
> is paramount for earning the trust of professors, since a fraction of
> them is used to Matlab/Scilab and a bit wary of Python (they will have
> the choice between Scilab and Python, although Python is recommended by
> the French "official program"): I cannot afford to take too many risks.
> What do you think? Any insights on using the notebook for teaching are
> very welcome.
>

0.13 can indeed open v2 notebooks, although it only saves in v3, so it's
not practical to use 0.12 and 0.13 together. But we intend to keep
compatibility for much longer now - the v3 format was designed to allow
much more extension without breaking the ability of 0.13 to read it. This
came up in the discussion around standardising the Scipy stack as well.
There we said that if we have to break compatibility again, we will need to
have an overlap period where IPython can read the new format, but still
saves in the old format by default.

Obviously the notebook is still quite young and rapidly developing, but we
see the format break from 0.12 to 0.13 as the exception, not the rule. If
you decide to use it, make sure everyone starts with at least 0.13.

That's great news about the scientific computing curriculum. Good luck with
training more people in Python!

Best wishes,
Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20121205/047c7098/attachment.html>