[Jprogramming] J on Jupyter progress

[Dan Bron]

Pascal wrote:
> Its easy to publish the results of an interactive session though.  github will preview them for example.
> 
> 
> https://github.com/adrian17/jkernel/blob/master/sample_edge_detect.ipynb
> 

After seeing this preview, I am *really* excited to see the final product here. I think being able to not only use J online but publish results from a notebook as documentation will be a killer app. It will really advance the foundational goal of communicating *with* J (as opposed to talking *about* J, or using J as a means to an end); I personally think this objective has been neglected in recent years.

Note the similarity of that notebook preview to the original J labs (which were, of course, limited to ASCII).  I can really see using this to answer someone on StackExchange, or someone asking you about the statistics of the presidential race, anything else where a bit of calculation is required. 

The big difference is because you can intersperse prose, media, and code freely, you can use J as no more than background, implementation details, scaffolding. Just use it to exhibit whatever you want to exhibit, without drawing attention to it as a notation. Which is exactly what a notation is supposed to do, and exactly the way to advertise a language which aspires to be “executable mathematical notation”. 

In other words: a way to sell J, without having to sell J (which, just like any other sale, makes people play devil’s advocate at least initially, purely as a self-defense mechanism).

-Dan

PS:  Having said all that, I personally have not done much to advance the cause of J in recent years. The problem with J is that it’s such a fun toy that it becomes a serious time sink.