On Sunday, 19 August 2018 at 20:33:45 UTC, Laeeth Isharc wrote:
Proof of concept works, but it requires some further
development to be useful to do work in.
https://github.com/kaleidicassociates/jupyterd
It uses D repl currently - this was written for a console
interface and probably you will encounter difficulties running
it in a notebook environment. I guess one would like to treat
all functions defined in a single notebook as part of the same
session and to execute immediate statements as part of a main
specific to that cell.
The kernel is a bit flakey - takes time to come on line and you
might need to reconnect to it sometimes.
To Do:
1. Add HTML and markdown table output to display arrays of
structs or of dicts in a useful manner
2. Integrate with mir and other numeric libraries
3. Integrate with charting
4. Consider adding to Dlang tour and run.dlang.io when stable
5. Integrate with dpp
6. Integrate with dub
1 and 3 should be quite simple. One wouldn't want to write a
large program in Jupyter, but it's helpful for exploratory data
analysis and programming where the code that does the work is
already in D.
Note that the D repl will only work on platforms where drepl
works i.e. platform with shared library support. It will _build_
on OSX due to
https://github.com/kaleidicassociates/jupyterd/blob/master/source/jupyterd/kernel.d#L393 but it won't work.