Flask application problem on Jupyter notebook installed through Anaconda

[ssarkaray...@gmail.com]

Hello,

I have a python application using Flask.  It runs fine from command prompt on Windows 10.

But when I try to run the same application from Jupyter Notebook (from Anaconda), I get following error for the

call :   app.run(debug=True, port=8000)

The error I get: 

UnsupportedOperation                      Traceback (most recent call last)
<ipython-input-13-b728e01956cc> in <module>()
----> 1 app.run(debug=True, port=8000)

~\AppData\Local\conda\conda\envs\deeplearning\lib\site-packages\flask\app.py in run(self, host, port, debug, load_dotenv, **options)
    936         options.setdefault('threaded', True)
    937
--> 938         cli.show_server_banner(self.env, self.debug, self.name, False)
    939
    940         from werkzeug.serving import run_simple

~\AppData\Local\conda\conda\envs\deeplearning\lib\site-packages\flask\cli.py in show_server_banner(env, debug, app_import_path, eager_loading)
    627             message += ' (lazy loading)'
    628
--> 629         click.echo(message)
    630
    631     click.echo(' * Environment: {0}'.format(env))

~\AppData\Local\conda\conda\envs\deeplearning\lib\site-packages\click\utils.py in echo(message, file, nl, err, color)
    257
    258     if message:
--> 259         file.write(message)
    260     file.flush()
    261

UnsupportedOperation: not writable Can I get some help Please ? Thanks, S. Sarkar

[Roland Weber]

You cannot run a permanent server inside a notebook kernel. Or more precisely, it doesn't make sense. The kernel is there to execute code snippets from a notebook or other client and return the output and results, not to permanently run a server application that listens on a different port. If you want to run a permanent server that listens on its own port, then start it from the command line. The notebook kernel doesn't add any value for that use case.

[Shaurya Khurana]

This is an internal issue of Jupyter Notebook, as it doesn't support this functionality. However, you can run this manually:

(Append this code at the end)

# starting the app on port

try: 

    # works fine while running the py script in the command-line 

    app.run(debug=True,port=8000)

except:

    # internal issue with Jupyter Notebook

    print("Exception occured!")

    # running manually

    from werkzeug.serving import run_simple

    run_simple('localhost', 9000, app)