2

I spent a week debugging a problem that got Python to crash (not give a scripting error, literally crash its VM) due to a random error.

After tracking this for days using procdump, I found that PySide 1.2.4 is causing the problem. Is their stable version unstable?

Dimitris Fasarakis Hilliard
  • 150,925
  • 31
  • 268
  • 253
The Quantum Physicist
  • 24,987
  • 19
  • 103
  • 189
  • @TheQuantumPhysicist I have attempted to give a facelift to this Q to maybe make it look like not much of a rant. Gil and piet are right, you need to make this look like less of a complaint and more like a standard Question and Answer. – Dimitris Fasarakis Hilliard Jul 06 '16 at 12:16
  • @Jim Thanks. I don't want this to be a rant. I want it to be helpful to others, because it takes a while to detect such bugs. – The Quantum Physicist Jul 06 '16 at 12:16
  • @TheQuantumPhysicist I understand and it's good that you have that mentality. You just need to package it in a way that is acceptable for StackOverflow :-) – Dimitris Fasarakis Hilliard Jul 06 '16 at 12:18
  • 1
    After discussion, I've withdrawn my vote to close. – Gil Hamilton Jul 06 '16 at 12:18
  • 1
    @Jim Agreed! Perhaps I'm subconsciously upset because of this. But I want this to help :-) – The Quantum Physicist Jul 06 '16 at 12:19
  • 1
    Like Gil, after the edit this is a better fit for the site. @TheQuantumPhysicist note that making a post community wiki is meant to make it easier to edit for other users but is not meant to make it ignore all SO-standards. – piet.t Jul 06 '16 at 12:22
  • This would be more useful if there was python code that would reproduce the problem. PySide wheel packages v1.2.2 and v1.2.4 have different Qt versions, v4.8.5 and v4.8.7 respectively, so there might be other factors. – J.J. Hakala Jul 06 '16 at 13:26
  • @J.J.Hakala The program I have is quite huge, and all I can say is that `QThread` is causing the problem from the dump (as you see in the screenshot). Sorry, but this is all I have. It's, however, useful to know that staying with 1.2.2 of pypi is error-free. – The Quantum Physicist Jul 06 '16 at 13:34

1 Answers1

4

According to pip standards, a stable release has to be put when a normal installation command is given, hence

pip install pyside

should provide a stable version. I spent a week debugging a problem that got Python to crash (not give a scripting error, literally crash its VM) due to a random error. After tracking this for days using procdump, I found that PySide is causing the problem. Following is the bug extracted from the dump file using WinDbg:

enter image description here

I went there to report the issue, and got shocked to find that the current standard version provided by pip/pypi (1.2.4) is not even stable! It's the dev version!!! Check this screenshot from their reporting website:

enter image description here

Of course PySide is not for free for commercial purposes, so are they using people's resources to test their dev version without telling them? This is so unethical and I don't want others to fall in the same trap I had. Please be aware of this!

Hence, when you install PySide with pip, use this command:

pip install pyside==1.2.2

Version 1.2.2 seems fairly stable with no such problems. Don't forget to uninstall the current version of PySide before installing the new one using that command.

Hope this helps.

The Quantum Physicist
  • 24,987
  • 19
  • 103
  • 189