Python 3.6 causes error, python 3.5 does not.

Jim jf_byrnes at comcast.net
Mon May 21 17:06:00 EDT 2018


On 05/20/2018 02:03 PM, Jim wrote:
> Mint 18
> Libreoffice 5.1.6.2
> Python 3.6.5 in one virtual environment
> Python 3.5.2 in another
> 
> I am writing a script that uses pyautogui to get some data and paste it 
> into a Libreoffice calc file, there by bypassing the complexity of uno.
> 
> The problem is it runs fine if I use python 3.5. If I use python 3.6 it 
> opens the calc file then pops up a dialog saying "std::bad_alloc". There 
> are no relevant errors in the terminal. At this point I must reload the 
> file and let calc recover it. I googled "std::bad_alloc" but didn'
> 
> This is the portion of the code that causes the error:
> 
> import pyautogui
> import subprocess
> import threading
> import time
> 
> LONG_ENOUGH = 5
> 
> class LO():
>      '''Manipulate libreoffice not using UNO'''
>      def __init__(self):
>          self.filename = 
> '/home/jfb/Documents/Financial/Investing/Stocks.ods'
>          opener = threading.Thread(target=self.open_it, 
> args=(self.filename,))
>          opener.start()
>          time.sleep(5) #LONG_ENOUGH)
>          #self.manipulate_LO()
> 
>      #Open and work with libreoffice
>      def open_it(self, filename):
>          #subprocess.call(["libreoffice", filename])
>          subprocess.run(["libreoffice", filename])
> 
> lo = LO()
> 
> To complicate matters not all .ods files show this problem. I ran some 
> tests. Two of my .ods files are effected but a couple of others are not. 
> A new test file I created is not effected.
> 
> I don't know if I have a Libreoffice problem, file corruption or a 
> Python problem. The fact that 3.6 gives an error but 3.5 does not is the 
> reason I decided to ask here first.
> 
> Regards,  Jim
> 

To follow up. I just discovered the error only happens if I start the 
script from inside the 3.6 virtual environment. If I start it from 
another terminal with the full path to the 3.6 VE and the script it runs 
without error.

Regards,  Jim




More information about the Python-list mailing list