Manual play not working

[marshallgrimmett]

Hi there,

I am having a problem getting the manual play to work.

I get the board to render, but when I click on a column it just says "processing..." and nothing ever happens.

Here's the code:

from kaggle_environments import evaluate, make, utils
env = make("connectx", debug=True)
env.play([None, "negamax"], width=800, height=600)

I have tried using "random" instead of "negamax" and I have also tried switching to second player, but nothing works.

Here are the versions I am using.

Python                        3.7.4
kaggle-environments           1.0.13
jupyterlab                    2.1.4
ipython                       7.13.0

Thank you,
Marshall

[harrisse]

Thanks for the report, we'll take a look!

[richard-hajek]

I second this. Reproduction is very easy. This docker-compose.yml:

version: '2.0'
services:
  jupyter:
    image: gcr.io/kaggle-images/python
    command: jupyter lab --allow-root
    volumes:
    - .:/working
    working_dir: /working
    network_mode: "host"
    stdin_open: true
    tty: true
    environment:
        HOME: '/working'

Browser: Firefox 80.0.1 (same result on Google Chrome 85.0.4183.102 (Official Build) (64-bit) )

Result:

And this just hangs like that seemingly forever

[SiggyF]

I have the same issue. I see the following error in the console log:

Error: IPython Unavailable: TypeError: Cannot read property 'notebook' of undefined
    at about:srcdoc:1171
    at new Promise (<anonymous>)
    at App.contextRef.current.execute (about:srcdoc:1158)
    at App.contextRef.current.act (about:srcdoc:1179)
    at HTMLCanvasElement.<anonymous> (about:srcdoc:332)
(anonymous) @ about:srcdoc:1199
Promise.catch (async)
App.contextRef.current.act @ about:srcdoc:1199
(anonymous) @ about:srcdoc:332

[bovard]

I've gone a number of reports on this on the kaggle forums as well. I will try to prioritize to take a look at this.

[SiggyF]

The current player code seems to depend on the ipython javascript API. I inspected the kaggle javascript environment using a javascript cell:

%%javascript
// Show python related variables
console.log('vars', Object.keys(window).filter(x => x.toLowerCase().match(/py/)))

This shows that the current kaggle version is using the JupyterLab environment. I couldn't find an easy way (other then by writing a jupyter lab extension to recreate the similar functionality. See also the open issue here .

[FelipeMaldonado]

I've gone a number of reports on this on the kaggle forums as well. I will try to prioritize to take a look at this.

Is there any progress on this? at least in my case the manual play works if I run locally a Jupyter notebook, but gets stuck in the 'processing' if I run it for example on Google Colab.

[jonahw23]

I'm having this issue as well, I've tried running a Jupyter notebook locally as well as using Google Colab and the Kaggle Notebooks, but I get the same "Processing" message no matter the enviroment. Any workarounds?

[kafkasl]

I'm having this issue as well, I've tried running a Jupyter notebook locally as well as using Google Colab and the Kaggle Notebooks, but I get the same "Processing" message no matter the enviroment. Any workarounds?

hey it didn't work for me using Jupyter Lab but it did work using normal notebooks.
In the terminal, instead of runningjupyter lab, run jupyter notebook