[IPython-User] data and code IO with notebook

Tim Michelsen timmichelsen@gmx-topmail...
Thu Sep 1 03:08:04 CDT 2011



>> Is there on option to have a storage area for code snippets (e.g. module
>> imports, functions) incorporated into the left panel of notebook?
> 
> I am not quite clear on what you are asking for and how it related to
> np.loadtxt.  The %run magic does work in the notebook and there is
> also the %loadpy magic which will import a .py file or URL into the
> notebook as a cell.  What are you wanting to have in the left panel
> exactly?  

> Can you describe the usage case more fully?
There are 2 types of use cases that would need data IO:

1) Working with existing python code such as a
	a) script stored on file on HD or
 	b) a fully programmed application stored on the same server
	c) code from a previous session in the notebook that one wants to pick up

   This is currently all only supported by %loadpy
   For item 1 above, the user would need to upload the file to a server
& load from there

   User expectations:
   User/group of users can open their notebook (maybe secured by log-in)
and find the history of the previous session, stored snippet (like in
%save), a list of uploaded code, a link to a application with which one
would work interactively. Such code could internally be administered by
a DVCS.

2) Working with data stored on disk or elsewhere that represents the
input for the calculations

  Again, such data needs to be uploaded and loaded from a separate server.

  User expectations:
  a button for the usual "Upload..." dialogue found in webapp like
email, a data storage that would store inputs and also outputs, a
procedure to export results (files, plots) to disk or to a static website.

This user solutions would offer the possibility move code development
and data analysis fully to the web opening new perspecitves for teams.
And by this way, colleagues who do not know how to program in Python but
are instructed to run a certain procedure in order to  analyse data
could benefit from scripts and code.

Maybe my ideas go beyond the scope of the notebook of which the
development has just started. My ideas may be more comparable to the
PythonAnywhere initiative. But chances are that IPython can be the
integrating brick for scientifically capable web apps.

Kind regards,
Timmie



More information about the IPython-User mailing list