Re: interest for PARI/GP Jupyter notebook ?

I have used Jeroen's pari kernel, though not recently. It is a very good platform for doing any kind of demo, for example of new pari/gp capabilities at an atelier, since you can have text cells as well as code cells so that's all you need for your talk.

I am not a good person to help make this work though, since I do not have the necessary expertise. When I just started up jupyter-notebook on my laptop it was not offering me a pari kernel, though I certainly did once have it working on the same machie, and I have completely forgotten what to do to install an extra kernel.

I have not tried Edgar's kernel -- from what Pascal said that might be easier.

John

On Fri, 12 May 2023 at 15:34, Pascal Molin <molin@math.univ-paris-diderot.fr> wrote:

I started using pari-jupyter two months ago, and immediately switched to Edgar's version because I had problems with the main pari-jupyter
(I do not remember precisely the issues, maybe it was slow or did not support threads). I also think a xeus version would be nice.
Anyway I think I will use it on a regular basis now and could provide feedback from myself and from students, and a bit of testing.

Pascal

Le ven. 12 mai 2023 à 15:34, Edgar Costa <edgarc@mit.edu> a écrit :
Hi Bill,

I strongly believe that having a jupyter kernel lows the barrier of entry to almost any language, and thus having one is worthwhile.
I am not a regular gp/pari user to help you with this one. I rarely write any gp code, I usually call it from Magma/Sage/Bash.
However, whenever I need to write more than one line command, I do it via some jupyter kernel, as I never remember the syntax, and this is a quick way to figure out what works.

The Xeus version is quite tempting, as it is much more natural for gp/pari.
However, cypari issues should get fixed regardless, as these have broader impacts.

Best,
Edgar

On Fri, May 12, 2023 at 8:41 AM Bill Allombert <Bill.Allombert@math.u-bordeaux.fr> wrote:
Dear PARI developpers,

Some time ago, Jeroen Demeyer has written a PARI jupyter kernel which is now
maintained mainly by Vincent Delecroix.
<https://github.com/sagemath/pari-jupyter>
this allows to create Jupyter notebook that uses PARI/GP.

However I do not think Vincent is actually using it, and I know nobody
that really use it, so I do not know how good is the user experience.

This kernel is based on python and cypari which causes some technical issues.

There have been two technically simpler alternatives, one using the GP binary
directly (by Edgar Costa) and one using Xeus <https://github.com/jupyter-xeus>
and libpari.

However before pursuing this further, we need someone that is willing to use
this kernel for actual applications, giving feedback and willing to test new
version which compiling and installing them.

So am I looking for someone interested to help.

Cheers,
Bill.