Hi,
On 26/01/17 05:43, Serge Stinckwich wrote:
On Wed, Jan 25, 2017 at 10:29 PM, Offray Vladimir Luna Cárdenas
<offray.l...@mutabit.com> wrote:
Hi Jose,
I'm building something that has ideas borrowed from Jupyter notebook[a], Leo
Editor[b] and Pharo/Smalltalk and few other original ones. The more I see
the connections between stuff like Eve[1], org-mode[2], Jupyter Lab[3], I
think that the time for literate computing[4] (a development beyond literal
programming), reproducible research and live coding is coming.
[a] http://jupyter.org/
[b] http://leoeditor.com/
[1]
https://hackernoon.com/smalltalk-and-protein-programming-4da245ac93e2#.2riwbeeia
[2] https://www.jstatsoft.org/article/view/v046i03
[3] http://jupyterlab.github.io/jupyterlab/
[4]
http://blog.jupyter.org/2015/07/07/project-jupyter-computational-narratives-as-the-engine-of-collaborative-data-science/
I'm currently using Jupyter notebook with some of my students for
doing simulation of dynamical systems and they love it I guess. I'm
not a big fan of the UI because it still looks like a web page and the
interaction is still somewhat limited compared to what you can do with
Pharo.
I'm also a big fan of org-mode and even more powefull than Jupyter
notebooks because you can mix different computer languages at the same
time and there is a real support for folding.
Yes, I am not a big fan of Jupyter notebook neither. Sounds kind of
mean, but the actual interface is like a glorified REPL log, powered by
HTML output, with a linear sequence of single input/output pair of
cells. So notebooks authors have two options: split the document into
several files, loosing the general overview of the document or a single
long "reel" of REPL log, with too much detail at the same time (I have
experience this by myself [1]). I think that org-mode and interactive
outliners (as exemplified in [2][3]) are better suited to manage
exploratory literate computing, interactive documentation, and the
"emergent" order that writing implies.
[1] http://mutabit.com/repos.fossil/piamed/doc/tip/Afiche/narrativa.png
[2] https://youtu.be/dljNabciEGg
[3] http://xiki.org/screencasts/
If you remember I have
done a small show us "your project" last ESUG about some experience I
have done about be able to run Pharo code inside org-mode.
Yes, I remember :-). I was also the "USB guy" in the orange cap team
:-P, so I was taking care of logistics and didn't see as much as I would
like of the event :-/
At the moment, we are quite interested in my research group to be able
to a kind of literate style of computing for modeling and simulation
of complex systems like simulation of epidemiology models. This is
something we can do with Jupyter but would be even more interesting in
the context of Pharo.
Eve looks also very interesting but I have to read more about this one.
[...]
I really like what you done until now with Grafoscopio and I will try
to help and use it for own work in the future.
Because of the uniform, moldable and integrated nature of Pharo, I think
that Grafoscopio, even in its early stages, prototypes stuff that
others, like Jupyter Lab, are trying to reach, for example trying to go
beyond the notebook and having a extensible computing environment
available to the notebooks. With Grafoscopio we already have that
(thanks to being "inside" Pharo).
Maybe a first try to integrate our work could be to have some Kendrick
notebook in Grafoscopio. I have to fix some important bugs, but
Grafoscopio is already usable for this kind of scenario and, of course,
any help is more that welcomed!
(There is still a nasty bug about playgrounds that produce text output
replacing the playground code, but I hope to fix it soon).
I have to look to Fossil. Is it possible to linked in someways to git ?
Yes, there is some fossil to git bridge. I will try to give support to
fossil first, to have a smooth integrated experience, but that doesn't
preclude the use of other DVCS to work with.
I was thinking that maybe that I will submit a gsoc proposal to have
some support in Jupyter notebooks for Pharo.
Apparently this not that difficult to built a basic support and after
that we can built some elaborate on top of it.
What do you about that ?
Regards,
I would like better a GSoC proposal to improve Grafoscopio, as a
literate computing environment for Pharo. Jupyter is popular but pretty
limited in terms of the notebooks document model and the moldabiliy of
the environment or the live coding capabilities (but it's improving).
For me, the big advantage of Jupyter or Org-mode is their language
agnostic architecture, while Grafoscopio is focused on Pharo/Smalltalk.
But language specific computing suites (like RStudio) are still
appealing for projects like GSoC.
A Grafoscopio GSoC proposal would try to improve several items:
- User interface and functionality: Keyboard shorcuts, GUI bugs,
integrated pdf export (via pandoc/LaTeX), integrated asynchronous
lightweight collaboration on documents (via fossil DVCS).
- Better test coverage and improving code quality in general.
GSoC rules doesn't allow being a mentor and a student, so I can not
mentor myself (despite of me kind of doing so with this project :-P). I
would really like to have someone mentoring me this project, helping me
with my knowledge gaps, Spec specific issues, coding in general.
So, would anyone of the seasoned Pharoers like to mentor this project?
I'm pretty self motivated, but having some guide for this summer to
improve my coding skills with a valuable project for the community will
make a big difference.
Cheers,
Offray