JupyterLab (beta) is out. Anyone tried it?


#1

Maybe of interest to the community: The more IDE-like JupyterLab was just released in beta: https://blog.jupyter.org/jupyterlab-is-ready-for-users-5a6f039b8906 (and HN thread). Looks pretty cool. As far as I understand it, it should just work with IJulia. Has anyone tried it?


Julia - IDE
#2

Very cool! I just installed it and am now playing around with it. it seems to work fine so far.


#3

I’ve tried it and I like it!

I also think that Julia is given good support. I can’t find much that it can do with python that it can’t do in Julia, which is good since its still in beta.

It has a super responsive data viewer for importing .csv files, but you can’t have the same viewer for dataframe objects. This is a functionality that Julia, and I think also python, lack, but is super helpful for data exploration, like R’s View() function and Stata’s browse function. So I hope something gets implemented to that effect.


#4

The tab-autocompletion doesn’t work for Julia. It works for python, though.


#5

That’s https://github.com/jupyterlab/jupyterlab/issues/3705 and should be resolved in next release.


#6

I spent most of today using it, and it’s great. The missing tab completion is annoying, but I’m sure it’ll be resolved soon. Being able to pin a cell’s output to a new pane is surprisingly excellent.


#7

It is excellent. One more quality of life improvement is I can select the kernel in jupyterlab rather than typing a rather hard to remember name on the command line for the notebook or qtconsole.


#8

In the standard Notebook you can select the kernel in Kernel -> Change kernel.


#9

Tab completion fix has been merged to IJulia master with support for displaying types in Jupyterlab:

I also just made a new pull request to IJulia that I think improves tab completion for symbols: https://github.com/JuliaLang/IJulia.jl/pull/634


#10

I was looking into this today, I’ve never been big on Jupyter notebooks, but I’m more and more starting to see the value in them from a collaboration/(get a nice document at the end of the workday) point of view. However, I was wondering if it was possible to directly inject code from a julia file into a running notebook. I ask this because I like to reuse a lot of code that I make, and so I sort of develop the package capabilities as I am working. My workflow now consists of using Atom to have one script file, and then when I want to add/alter a function to the package I just quickly evaluate the function definition in the respective file/module. I’m not sure if that can be mimicked using notebooks and Revise.jl, without actually copy pasting the code left and right.


#11

Couldn’t you just use include from the notebook (in combination with Revise.jl)?


#12

In the console mode I’ve found that just arrows up and down for previous entries means i don’t have to copy and paste very much. just “up, up, up, shift-enter”


#13

Yea I think I underestimated the power of Revise.jl, last time I looked at it was over a year ago. It seems that now it should be able to accomplish what I want to do.


#14

Does anyone happen to know if it is possible to run a kernel on the server, but with the packages etc from the local machine? I managed to get a regular notebook server running to which I can then connect and start notebooks, but that’s not really what I want. I would like to connect local notebooks to the server. I found this jupyter_kernel_gateway thing but I have no clue how to configure it and can’t find too many references online.
In their docs, there is a Usecases section stating:
“Attach a local Jupyter Notebook server to a compute cluster in the cloud running near big data (e.g., interactive gateway to Spark)”
Which sounds exactly what I want to do. Any ideas?