Announce of the workshop on kotlin jupyter plugin building: https://kotlinlang.slack.com/archives/C4W52CFEZ/p1621060869158800?thread_ts=1620221118.142100&cid=C4W52CFEZ cc @Ilya Muradyan

I'm trying to create a Datelore Kotlin notebook with a custom Maven Repo using @file:Repository command (it is a JFrog based Repo). However when it then tries to resolve a module (I'm using @file:DependsOn), it says it cannot find that module. Strange thing is however that the first line it said it tried, is actually the correct URL. And this URL does resolve fine when entered in a browser. So the URL it looks for is correct but somehow still cannot retrieve it. Does anyone know if Datalore has some restrictions on connecting to outside world (like egress rules) or perhaps doesn't play nice with some of the redirection that sometimes goes one with Maven Repo's ? Not sure this is best channel to ask this question, but seemed the most appropriate one.

Just succeeded with deploying the Kotlin notebook with DataLore. In case anyone want to see it (very small demo of what package can actually do): https://datalore.jetbrains.com/view/notebook/umJ5HakN3gpKcI1VDbKw52

@Peter The problem is that in Plotly.kt I can't properly load the JS bundle before the execution because notebook isolates different cells. How do you manage to do that in your example? Isolated pages + js from cdn?

Thanks to inspiration from @Peter Plotly.kt v0.4.2 is released with a legacy notebook trigger! Now it works in classic notebook and in the DataLore!

Hi, could anybody suggest something resembling numpy's

meshgrid()

and

multivariateNormal()

in the current Kotlin ecosystem? I'm looking to rewrite code in these notebooks without `kotlin-numpy`: https://nbviewer.jupyter.org/github/JetBrains/lets-plot-kotlin/blob/master/docs/examples/jupyter-notebooks/contours.ipynb https://nbviewer.jupyter.org/github/JetBrains/lets-plot-kotlin/blob/master/docs/examples/jupyter-notebooks/density_2d.ipynb

I am playing a little bit with kotlin-jupyter and I am getting exception after simple

typealias Matrix = List<List<Double>>

Crash is in thread. It is of course working if I define it as a class.

in krangl is there a way to quickly combine multiple dataframes into a big one?

Playing around a bit with Datalore and Kotlin notebooks. Have to say, a very nice and simple environment. But I noticed that on a basis 4GB machine, the Java max heap size is only 1GB. Is there anyway to influence this?

Some time ago I promised a Proof Of Concept implementation of widgets for kotlin jupyter kernel. Today I finally had some time to work with it. Here it is: https://github.com/altavir/jupyter-mpp-demo. No documentation yet, but there is a demonstration notebook included. Currently only checkbox and text field are implemented and there are some rough edges ( for example only one instance of the same widget is connected), but it seem to be alright for POC. The architecture is extensible, so it is possible to add new widgets quite easily. If someone wants, it could be improved to the full-scale widgets library. Also you can use this example as a template for your own full-stack plugins. As always, thanks to @Ilya Muradyan for fantastic cooperation.

Quick question regarding the kernel: Is there a keyboard shortcut to show method arguments help? Like Ctrl-P in the IDE

@Ilya Muradyan Not sure if this is intended or a bug, but in the IDE a smart-cast works that fails in jupyter:

I've finally managed to implement

@JupyterLibrary

support for

krangl

and

kravis

in https://github.com/holgerbrandl/krangl/blob/master/CHANGES.md#v017 and https://github.com/holgerbrandl/kravis/blob/master/CHANGES.md#v08. I took the opportunity to improve krangl-interop with

lets-plot

and have rebuilt one of my favourite R/ggplot teaching examples in kotlin&jupyter. Enjoy https://nbviewer.jupyter.org/github/holgerbrandl/krangl/blob/master/examples/jupyter/sleep_patterns.ipynb. Feel welcome with feedback and criticism. @Ilya Muradyan What is the intended process for the jsons such as https://github.com/Kotlin/kotlin-jupyter/blob/master/libraries/krangl.json? Are they legacy and will be kept, or will they be removed as soon as a library adopts the new integration model?

I was so excited today when installing 2021.2 to finally explore https://plugins.jetbrains.com/plugin/16340-kotlin-notebook . It installs, and it actually run code-cells within the IDE horay, but code analysis seems broken for me for kotlin and markdown cells. The error is shown in the screen shot. What am I doing wrong?

Is there an API to limit the output of a Jupyter Kotlin kernel. So the equivalent to "%logLevel off" but then to be called from within the kernel?

I'm looking for some kotlin alternative to plot on a map (like folium with python)

@Ilya Muradyan We just have a problem with kapt. Again. I've raised https://github.com/Kotlin/kotlin-jupyter/issues/325. I think it makes sense not to use it by default.

Together with Andrei Kislitsin we have developped a platform demonstrating the use of coroutines for reactive streams, KMath and Plotly.kt for high frequency trading analytics. Keep an eye on the repo as we will continue to add more and more interesting examples especially involving our Deep Learning framework within KMath based on pytorch: https://github.com/grinisrit/noa-atra-examples

I'm sorry for the dumb question, but how can I load a csv file in datalore kotlin?

DataFrame.readCsv from krangl would do it

Plotly looks working fine within Datalore.. (

%use plotly

) If I would like to render a graph like that, what syntax/code shall I use? (Sorry for the dumb question)

How to get the GPU instance (which is adverised to be free for 10h p.m.) on datalore?

is there any util that helps printing aligned all the fields of the classes inside a collection? Sort of like

DataFrame

@Ilya Muradyan Do we have a library plugin for jupyter which loads coroutines? I see there is one for serialization.

In Kotlin Notebooks, is there a way to catch exceptions before they are displayed in the output of a cell so they can be formatted a bit nicer?

Have been playing around with the DataFrame library, very nice and flexible and also great progress since last time I looked at it. But I have a few questions about the rendering, so if there are any experts that can enlighten me: • It looks/behaves very different when displaying a table between Datalore and Jupyter. Is this due to Datalore specific capabilities? • It contains some hard-code CSS styles that makes it not play nice with Jupyter Lab themes. Without these any additional CSS rules it seems to render much better. Can DataFrame be configured to not include any custom styles or overwrite them?

Is the Kotlin kernel providing its own version of kotlin-jupyter-api jar and put this also first on the classpath? The reason I ask is that I wanted to try out the latest version of kotlin-jupyter-api (since it has a nice new feature) and my code compiles fine. I also see that version in the local maven cache, so all good. But as soon as I try to load this in a notebook it fails and the reason is that somehow an older version of kotlin-jupyter-api is loaded instead.

I have a strange requirement: end users of a notebook are required to add a single jar file to the Java class path. This is because the company that provides that jar file, only allows the end-users to use and distribute that jar file and I cannot package it any other way. So my current way of thinking: Step 1 User uploads the jar file to some Jupyter Notebook directory Step 2 Somehow that jar file gets added to the Java classpath. Could perhaps be a function the user could call from a cell, for example: addCustomJar("propriatary-api.jar") • This then could be directly added, using some URLClassLoader magic in the back ground • or install it in the local maven repo (~/.m2) But open to alternatives and any suggestions. Or perhaps something I overlooked that is already available out of the box. Any insights appreciated!

Im looking for a data structure that is basically a

double[10_000][300]

with fast

contains

and fast

indexOf

, any suggestions?