MobileRead Forums - View Single Post - Python function terminated unexpectedly...

Kindle3w_Pdfs · 12-10-2017, 12:58 PM

@Kovad: I don't believe I asked you to prove a negative that's impossible to prove. I merely asked whether you know the internals of the Python functions called (directly or indirectly) by Calibre well enough to be certain that Python doesn't modify those Registry values.

Regarding the different Comodo behavior for Calibre and Calibre-debug, it should not be attributed to a bug in Comodo. I had a custom Comodo HIPS rule set up for Calibre but not for Calibre-debug. Note: The custom rule didn't prevent an earlier version of Calibre from working properly, and when I removed the custom rule Calibre 3.13 ran okay.

It seems reasonably likely, from my perspective, that an update of a third party library (probably Python) is behaving in a way you don't fully understand, and that you're blaming users' antivirus software in such cases. Perhaps you're more familiar with the internals of an older version of Python that didn't behave this way; I wouldn't expect you to spend time studying the details of every update made to Python. I don't mean to give offense, and if I'm wrong I hope you can appreciate that this theory is plausible from my perspective.

Question: In which Calibre source code module(s) is the code involved in accessing the internet during Calibre startup? Since Calibre is open source, maybe I can find the problem, if I know where to start looking.

@DoctorOhh: Whether or not changes to the Registry are logically 'required' (to collect usage statistics) is irrelevant. What is relevant (to the antivirus defense) are: (1) Does the app makes changes to the Registry? (2) Does the app ask the operating system to grant a privilege that would allow the app to make Registry changes later? The Python function that issued the error message when blocked by Comodo may have requested more access to the Registry than is really needed to send usage statistics. See for example: https://stackoverflow.com/questions/...-access-denied (where the programmer may have inadvertently used a Registry access parameter that asks for an unnecessarily high privilege.)

12-10-2017, 12:58 PM	#22
Kindle3w_Pdfs Enthusiast Posts: 27 Karma: 10 Join Date: Nov 2010 Device: kindle 3 wifi	@Kovad: I don't believe I asked you to prove a negative that's impossible to prove. I merely asked whether you know the internals of the Python functions called (directly or indirectly) by Calibre well enough to be certain that Python doesn't modify those Registry values. Regarding the different Comodo behavior for Calibre and Calibre-debug, it should not be attributed to a bug in Comodo. I had a custom Comodo HIPS rule set up for Calibre but not for Calibre-debug. Note: The custom rule didn't prevent an earlier version of Calibre from working properly, and when I removed the custom rule Calibre 3.13 ran okay. It seems reasonably likely, from my perspective, that an update of a third party library (probably Python) is behaving in a way you don't fully understand, and that you're blaming users' antivirus software in such cases. Perhaps you're more familiar with the internals of an older version of Python that didn't behave this way; I wouldn't expect you to spend time studying the details of every update made to Python. I don't mean to give offense, and if I'm wrong I hope you can appreciate that this theory is plausible from my perspective. Question: In which Calibre source code module(s) is the code involved in accessing the internet during Calibre startup? Since Calibre is open source, maybe I can find the problem, if I know where to start looking. @DoctorOhh: Whether or not changes to the Registry are logically 'required' (to collect usage statistics) is irrelevant. What is relevant (to the antivirus defense) are: (1) Does the app makes changes to the Registry? (2) Does the app ask the operating system to grant a privilege that would allow the app to make Registry changes later? The Python function that issued the error message when blocked by Comodo may have requested more access to the Registry than is really needed to send usage statistics. See for example: https://stackoverflow.com/questions/...-access-denied (where the programmer may have inadvertently used a Registry access parameter that asks for an unnecessarily high privilege.)