Thanks, Matthias.

Yes, https://github.com/lstagner/idl_kernel 
<https://www.google.com/url?q=https%3A%2F%2Fgithub.com%2Flstagner%2Fidl_kernel&sa=D&sntz=1&usg=AFQjCNFtWFr3bFcYPWz-MzlEC3oxOHgi7A>
 
is the page I was referring to. In it, there is a statement:
   #This package is no longer maintained since IDL now has an official 
Jupyter kernel <https://www.harrisgeospatial.com/docs/IDL_Kernel.html> 
#IDL/GDL kernel for IPython/Jupyter
and the link is as you described.

What I failed to realize is that IDL is a commercial software. Once it is 
bought and installed, perhaps there will be a IDL_DIR with the appropriate 
file.

Because of the syntax (mostly) compatibility between the two (IDL & GDL), I 
was under the impression that the same .py file can be used to link up 
Jupyter and GDL, with possibly minor modifications, but perhaps I am being 
too simple-minded about what a Jupyter kernel actually is. I have never 
seen a gdl_kernel.py, I was hoping one existed somewhere.

I do not know much about python, but since Jupyter is trying to be a 
language-agnostic tool, which I like very much and would like to use in my 
teaching, I sort of assumed that a way to link a pipe to an external 
executable that would receive my text and return a tagged (html?) stream, a 
mixture of text and svg graphics, would be a matter of adjusting a 
configuration file, mostly. Isn't this what Jupyter does when the cell 
magic is set to %%bash, for example?

The instructions for installing a new kernel that I have found out there 
are universally inadequate. Under RHEL 7, I have access to pip, but not 
conda. Unfortunately, the simple-minded 

  pip install gnuplot_kernel

to use an example, I have tried several, not one of them showing up under the 
list of Jupiter notebook kernels available. However, if I do this instead
  pip install --upgrade --no-cache-dir 
git+https://github.com/has2k1/gnuplot_kernel.git@master
it works as expected! I have not yet discovered the equivalent solutions for 
other kernels, or how to make my own (to an external executable... see above). 
  $ pip list | grep kernel
  gnuplot-kernel (0.2.3)
  ipykernel (4.6.1) 
  metakernel (0.20.8)
  octave-kernel (0.28.1)
  scilab-kernel (0.8.1)
yet the Jupiter kernel drop down menu only lists Python, Python 3, and octave.  
And, after I re-did the installation as above, gnuplot, but no scilab, for 
example.

My interest is to unite several different legacy tools under one Jupyter 
"roof", and I want to use multiple kernels in a single notebook, though I am 
having trouble with that 
right now: it seems that it's only possible if the "primary" kernel is python, 
because in my octave-kernel notebook, the command 
   %load_ext gnuplot-kernel 
does not seem to be working for me, presumably because octave kernel does not 
have the load_ext functionality, like a python kernel does. Before I start 
looking into
writing custom kernel interfaces to my other tools, I need to make sure I 
understand and can make robust the installation process.

The gnuplot trick I found here: 
   https://github.com/has2k1/gnuplot_kernel/issues/5
together with the expressions of puzzlement of why it was needed, so I am not 
the only one confused.



-- 
You received this message because you are subscribed to the Google Groups 
"Project Jupyter" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jupyter/a6f19fa7-c1a4-41b7-8da3-a825278a074a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to