[IPython-User] Canncial way to do web based ipython shell?

Anthony Oliver anthony@ingenuitas....
Thu Jan 19 13:30:45 CST 2012


No this helps a lot.  I want the functionality and have looked at a lot of
other apps but we have been using ipython before the notebook
implementation and feel it's way better than anything else out there.  I
spent some more time yesterday poking around the code.  I think I can pull
out what I don't want from it.  In reality I basically want all the same
functionality as the notebook, I'm even fine with keeping a history, etc.
 Basically I think I just need to tell it to use the same kernel every time
instead of firing up another instance.  Does this sound correct? Or would
the better way be to start a new kernel instance and kill any old ones

I guess the TL;DR is I want a singleton instance of a notebook. ;-)
The way it sounds it isn't possible with the way it's designed.  We are
currently using cherrypy, not sure what it would take to port over to this,
but we can use tornado if that is a big staple as well.

p.s. Thanks to the ipython team for all your work, anytime I jump on a
system without ipython now it makes me a sad panda.

Anthony Oliver
Chief Technology Officer
(906) 289-8169

On Wed, Jan 18, 2012 at 1:19 PM, Brian Granger <ellisonbg@gmail.com> wrote:

> On Tue, Jan 17, 2012 at 1:45 PM, Anthony Oliver <anthony@ingenuitas.com>
> wrote:
> > Is there a correct way to just build the shell that could work within a
> > browser, or an example somewhere?
> You would need to build a custom web application from the ground up.
> The best example is the current notebook code.  It shows how to use
> ZeroMQ to communicate with IPython kernels using JSON messages.  But
> the code is not really designed to be used as a library.  We plan on
> moving in that direction, but it will take time.
> > I basically want the functionality of the notebook.  My problem with
> using
> > notebooks is that we are currently using ipython as an embedded shell
> which
> > is awesome. But we want to extend it through a web interface, the
> notebook
> > method is perfectly fine, except in our application, the current running
> > program is always in context, so you can think of it as only ever needing
> > one notebook.
> >
> > I started diving into the notebook app itself, it appears much of the
> > javascript is also looking for things (for instance I tried to override
> the
> > handlers for the paths so the same thing is always served).  I can
> continue
> > on my path, but I would imagine there maybe a more stripped down example
> > floating around some where.
> Unfortunately we don't have any other examples.  On one hand, the
> notebook app is a fairly simple example itself.  It shows the basics
> of how you would communicate with kernels and display output, etc.  On
> the other hand, notebook web application is extremely complex.  There
> are multiple channels of communication (WebSockets, Ajax), everything
> is async and you have to wade through our JSON based message protocol.
>  Not sure if this helps, but I think your approach of diving into the
> notebook code base is a good one.
> Cheers,
> Brian
> > Thanks.
> >
> > ---------------------------------
> > Anthony Oliver
> > Chief Technology Officer
> > http://www.ingenuitas.com
> > (906) 289-8169
> > anthony@ingenuitas.com
> > ---------------------------------
> >
> >
> > _______________________________________________
> > IPython-User mailing list
> > IPython-User@scipy.org
> > http://mail.scipy.org/mailman/listinfo/ipython-user
> >
> --
> Brian E. Granger
> Cal Poly State University, San Luis Obispo
> bgranger@calpoly.edu and ellisonbg@gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/ipython-user/attachments/20120119/d31a73e4/attachment.html 

More information about the IPython-User mailing list