Kernel busy even after closing the browser

[contang0]

For some reason Python kernel continues to be busy even after I close the chrome tab that contains the dataframe I wanted to view. I basically have to restart the kernel every time I use tabloo.show because it does not stop automatically when I close the viewer.

Any way to solve this? If it helps, I'm on windows.

[bluenote10]

Tabloo tries to satisfy multiple use cases, not only local data visualization. It is designed as a web server, so it is more of a dashboard app, potentially allowing multiple users to access the visualization. Closing a browser window is not really something a web server can/should react to. A few ideas:

• We could add a button to the UI to shutdown the web server, only if the webserver gets launched from the convenience launcher tabloo.show(). For real dashboard use cases the shutdown option would be omitted.
• Add a possibility to gracefully shutdown the flask server from the terminal / backend process e.g. by a keypress.

I'll have to experiment what works best.

[bluenote10]

I have investigated the problem a little bit. By default the flask server was running in debug mode, which comes with an auto-reload functionality that basically relaunches the server when it gets killed.

I have released a new version where debug mode is disabled by default. This should now allow to simply press CTRL+C in the process that runs tabloo.show(). Let me know if this works better.

[contang0]

Thanks a lot for looking into this!

Ctrl + C simply kills the process, so the end result isn't much different from restarting the kernel. I'm working with a lot of dataframes and I inspect them after changing something. It takes a while to load all the data, so restarting the kernel after viewing a dataframe a single time isn't feasible for me. I think the option with the button in the UI makes the most sense.

[bluenote10]

Ctrl + C simply kills the process

Maybe I need a bit more details about your setup/workflow. I cannot reproduce this. For me CTRL+C only kills the flask server process, not the main process:

I'm working with a lot of dataframes and I inspect them after changing something.

In the long term I want to support visualizing multiple data frames, and an update interface would fit nicely. Maybe something like:

tabloo_server = tabloo.server({
    "Dataset 1": df1,
    "Dataset 2": df2,
})

# and later
tabloo_server.update({
    "Dataset 1": df1_modified,
})

[contang0]

I guess this is due to the fact that I'm on windows.

Normally I run code from Atom + Hydrogen, so ctrl + C would not work in that case anyways.

Visualizing multiple dfs sounds very nice. How about a side pane where the names of all currently defined dataframes would be shown, and clicking on each would show what it contains?

[ryanalarsen]

I'm coming across this issue when calling tabloo from a PyQT5 gui. Pressing the button launches tabloo, but afterwards, the kernal is busy and there seems to be no way to resume the program.

I've tried calling tabloo from inside subprocess and from inside a thread using the threading module. No luck as of yet.

[bluenote10]

I've tried calling tabloo from inside subprocess and from inside a thread using the threading module. No luck as of yet.

A few thoughts:

• Did you try disabling debug mode?
• Do you explicitly kill the subprocess yourself?
• Do you mean that it does not run properly from the subprocess, or does the stopping not work?

[jy247]

Could you add a kill button to the front-end? I am trying to call it in a similar way and also having issues with orphaned processes.

thanks