Hi Phil,

On 06/05/18 07:30, phil@highoctane.be wrote:
Hi Offray,

On Sun, May 6, 2018 at 1:47 PM, Offray Vladimir Luna C��rdenas <offray.luna@mutabit.com> wrote:

Hi Phil,


On 27/04/18 03:02, phil@highoctane.be wrote:

On Thu, Apr 26, 2018 at 3:57 PM, Offray Vladimir Luna C��rdenas <offray.luna@mutabit.com> wrote:
Hi,

I will be showcasing Pharo, Roassal and Grafoscopio at re:publica 2018,
next week. As you may know, re:publica[2] is one of the most important
and visible media & digital culture conventions in Europe and is a good
scenario for the Pharo community.�� You can find details about my
participation at [1]

The "on how to use the Grafoscopio��pocket infrastructure��for data activism and digital citizenship��" soundbite just blew my mind.

"Pocket Infrastructure" hits the nail on the head for sure.��
I positively love that angle.��
Compared to python, node, etc, there is much less fuss to get started and it allows to focus on the data narrative right away.

I regularly make folders in my Dropbox where I throw a pharo vm and a pharo image I work on, saved using��https://github.com/Pharophile/HOImageSaver so that I can always go back in time easily (saved me more than once!). This very handy for working on machine A, and resuming work on machine B, or C etc.
And is indeed "Pocket Infrastructure".


I think that pocket infrastructures are a powerful concept for inclusion and participation, specially in the times of fancy exclusionary "Big Data" and "Artificial Intelligence" buzzwords and discourses that don't take into account who becomes data of who, who will have the access, storage and processing capability in those forecast futures. In contrast Pocket Infrastructures are inclusive by default, being self-contained, simple and working well in on/off line contexts.

As I am busy in that space with Hadoop / TensorFlow / ..., yes there is a tendency to have kind of "high priests of data" showing up.
This is isolating the common folk instead of empowering them. I am fighting that tendency in the projects because the technology can be approachable.


Well, now we are two of us. In general the Smalltalk tradition of fighting against incidental complexity and the needs of broader communities of being empowered by code+data will bring more people interested that is already active on those concerns.



Pharo, in the live coding environment front and Fossil, in the DVCS front, work pretty well for that working definition of Pocket Infrastructures.�� Is impressive that just under 50 Mb anyone can have a Jupyter+GitHub alike environment for data storytelling, visualization and reproducible research that just run in their pockets and low end machines (and of course, big server and anything in between). We tried before with IPython notebook and other more complex stuff and we were dealing with external complexities instead of going right into story telling + coding + data viz. Of course, Jupyter and GitHub are a lot more popular that Grafoscopio + Fossil, but they're also traveling the most traverse path, while I think that a lot of valuable innovation comes, by definition, from the margins and that can give Pharo ecosystem an advantage point over other more popular approaches, as practice have demonstrated time and again.


As a Jupyter user, yes, I can relate. Jupyter is easy once you have it installed (like installing Anaconda distribution and typing "jupyter notebook" on the command line. But that is already a high bar for non technical people.

As a former regular user of IPython Notebook (before it morphed to Jupyter), my issues go beyond installation and are more related with moldability and the possibility to adapt the tool taking into account its users and contexts, hopefully by such users in such context. In the end, each Jupyter notebooks provides a RELP "linear metaphor" for documents and in that case you have two options, which I don't think are the best for exploratory computing, start to split your document into several notebooks (losing panoramic view) or ending with long scrolls, losing detail[1][2]. The best, for such exploratory documents, would be to combine the outliner metaphor of Leo/OrgMode with the interactivity of IPython[3], but in that endeavor you end fighting with a lot of incidental complexity that Pharo doesn't have[4], because of the way liveness, simplicity and adaptability are embedded into the tool and culture. In that way, Grafoscopio notebooks can contain simple notes or full books [4a][4b]

[1] http://mutabit.com/repos.fossil/piamed/artifact/d8762d656bcbf480
[2] http://mutabit.com/repos.fossil/piamed/doc/tip/linea-base-aras-iecas.html
[3] http://mutabit.com/offray/static/blog/output/posts/on-deepness-and-complexity-of-ipython-documents.html
[4] http://mutabit.com/offray/static/blog/output/posts/grafoscopio-idea-and-initial-progress.html
[4a] http://mutabit.com/repos.fossil/grafoscopio/dir?ci=28ec1bf0c1ab3c71&name=Docs/En/Books/Manual
[4b] http://mutabit.com/repos.fossil/mapeda/

Even with JupyterLab and the phospore.js[5] that improves on extensibility, you still need to deal with a lot of friction in the development experience.�� I would say that Grafoscopio and JupyterLab are following inverse paths (and in that way one is the future of the other): Grafoscopio started from the Pharo Interactive Development Environment and is trying to bring interactive notebooks documentation capabilities to such environment, while Jupyter started with interactive notebooks and is trying to become a full IDE [7][7a]. Grafoscopio will learn a lot by increasing languages support, and JupyterLab is dealing with development needs and customized workflows.

[5] https://github.com/jupyterlab/jupyterlab
[6] https://medium.com/@brianray_7981/jupyterlab-first-impressions-e6d70d8a175d
[7] https://youtu.be/Ejh0ftSjk6g?t=253
[7a] https://youtu.be/Ejh0ftSjk6g?t=329


I am very interested in Grafoscopio to work out proof of concepts in Pharo, so that I can have a nice narrative structure.
Coupled with gtExample and specific inspector presentations, this is a terrific thing.

I usually start from a notebook. There I put notes, problem descriptions, future working playgrounds where I launch the debugger. Is a really good experience and with your your proposed bug fix for the annoying UI bug is even getting better. It was the kind of experience I would like to have with Jupyter+Leo+Python, but now empowered by Pharo :-). Let me know how it works for you.


My current project involved a Glorp component and I'd like to showcase transactions within a Grafoscopio notebook.

I'll tell you how it goes :-)

I'm really curious. Keep us informed. :-).


Fossil is indeed a great vehicle for code and assets. Fossil and SQLite, great stuff.

Phil

Yes, good combination, even for the released open data of the biggest leak in journalism history, the Panama Papers, SQLite+Fossil+Pharo[8] worked pretty well. I think that there is a lot of unseen potential in such simple but powerful systems in all these places where people is just using popular "big data" stuff and we need to think in the "last mile" infrastructure to connect people with data. There the Pocket Infrastructures have a lot to offer and we can pioneer the path.

[8] http://mutabit.com/offray/blog/en/entry/panama-papers-1

Cheers,

Offray