halo
halo copied to clipboard
manrajgrover
iPython notebook keyboard interrupt breaks cell when using Halo
Description
Currently, in an iPython notebook, if a user does not handle keyboard interrupt and the spinner is running, a KeyboardInterrupt is thrown but the spinner keeps spinning.
System settings
- Operating System: MacOS
- Terminal in use: iTerm 2
- Python version: 2.7.10
- Halo version: 0.0.8
Error
Expected behaviour
In [5]: import time
...: import random
...:
...: spinner.start()
...: for i in range(100):
...: spinner.text = '{0}% Downloaded dataset.zip'.format(i)
...: time.sleep(random.random())
...: spinner.succeed('Downloaded dataset.zip')
...:
â ¹ 4% Downloaded dataset.zip^C---------------------------------------------------------------------------
KeyboardInterrupt Traceback (most recent call last)
<ipython-input-5-81a625f23371> in <module>()
5 for i in range(100):
6 spinner.text = '{0}% Downloaded dataset.zip'.format(i)
----> 7 time.sleep(random.random())
8 spinner.succeed('Downloaded dataset.zip')
9
KeyboardInterrupt:
â ¼ 4% Downloaded
Steps to recreate
Following snippet should recreate the error.
In [1]: from halo import Halo
In [2]: spinner = Halo(text='Downloading dataset.zip', spinner='dots')
In [3]: import time
...: import random
...:
...: spinner.start()
...: for i in range(100):
...: spinner.text = '{0}% Downloaded dataset.zip'.format(i)
...: time.sleep(random.random())
...: spinner.succeed('Downloaded dataset.zip')
People to notify
@ManrajGrover
In this case, if you type spinner.stop() (doesn't saw but it is kept handling) and press Enter, the spinner stops. But I think you want a more elegant solution.
To solve this problem, we capture KeyboardInterrupt signal, pass to the thread and stop a task immediately. One of the solutions is to implement specific thread class inherited thread.Thread like this. Second is to using signal module. In this way code is more clear and much better but there are maybe platform-specific issues.
I used the second way to solve the problem. Please review my method.
@JungWinter I would need a little more time to review the methods. Apologies for the delay.
With reference to #38, adding test case seems to fail on Appveyor. Need to find the reason behind this.