Bug 592 – shiboken.dll produces a segmentation fault when reloading a PySide module

PySide Bugzilla Closed for New Bugs

PySide is now a Qt Add-on and uses the Qt Project's JIRA Bug Tracker instead of this Bugzilla instance. This Bugzilla is left for reference purposes.

Bug 592 - shiboken.dll produces a segmentation fault when reloading a PySide module

Summary:

shiboken.dll produces a segmentation fault when reloading a PySide module

Status:	CLOSED WORKSFORME

Product:	PySide
Classification:	Unclassified
Component:	PySide
Version:	1.0.0 beta1
Platform:	PC MS Windows XP/Vista/7

Importance:	P2 normal
Assigned To:	renato filho

URL:

Depends on:
Blocks:
	Show dependency tree / graph

Reported:	2011-01-10 09:24 EET by Andreas Dewes
Modified:	2011-01-21 15:44 EET (History)
CC List:	8 users (show)

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.

Description Andreas Dewes 2011-01-10 09:24:45 EET

For the Windows XP binary (1.0.0beta1), shiboken.dll gives a segmentation fault
when I import a Python module that defines some PySide classes, reload it
afterwards and import it again. Example:

pyside_reload_test.py:
---
import sys

from pyside_module_test import *
reload(sys.modules['pyside_module_test'])
from pyside_module_test import *
---

pyside_module_test.py:
---
from PySide.QtGui import * 
from PySide.QtCore import *

class test(QWidget):
  pass
---

Running "python pyside_reload.test.py" produces a segmentation fault in
shiboken.dll. I'm running Windows XP 32 bit Service Pack 3 and Python 2.6

Comment 1 Matti Airas 2011-01-10 10:50:28 EET

Thanks for the bug report. I wasn't able to reproduce the bug with PySide 1.0.0
beta 3 on Ubuntu 10.10, so this might be already fixed, or then it's a issue
specific to Windows. In any case, I'm prioritizing the P3. This will be fixed
during the next sprint, at latest (in 4 weeks).

Comment 2 Hugo Parente Lima 2011-01-10 11:33:25 EET

Shiboken-based bindings doesn't support module reloading (yet), so I'm a bit
surprised that it doesn't seg fault on Linux too.

Comment 3 Matti Airas 2011-01-10 12:31:28 EET

Hugo, in your opinion, should I reprioritize to P4, to make this post-1.0
content?

Comment 4 Andreas Dewes 2011-01-10 13:30:27 EET

Thanks for your feedback, I guess this is a problem specific to Windows (or
even some special version of XP). I will recompile the library from source and
check if the bug is still there afterwards.

Comment 5 Hugo Parente Lima 2011-01-10 14:30:47 EET

Matti, IMO P3 or P2, as I said our generated code is not prepared for a module
reloading, so I'm a bit surprised that it doesn't crash on Linux too.

Comment 6 Matti Airas 2011-01-11 06:28:52 EET

OK. I thought support for reloading should've been regarded a new feature, and
that's why I was proposing postponing it. If not, I'm raising the priority to
P2 because it's a crasher issue, and (theoretically) not limited to a single
platform.

Comment 7 Marcelo Lira 2011-01-19 18:37:51 EET

Andreas, I turned your example into tests for Shiboken and PySide:

shiboken/tests/samplebinding/module_reload_test.py
shiboken/tests/samplebinding/test_module_template.py
pyside/tests/QtGui/pyside_reload_test.py
pyside/tests/QtGui/test_module_template.py

I've run them on Linux and Windows they worked for me.
I'll mark the bug as works for me, if you run the tests and the problem persist
please reopen it with any new information that arises from running the tests.

For information sake here's a link from Python bugs telling that proper reload
for C extension modules is not supported:
http://bugs.python.org/issue1144263

Comment 8 renato filho 2011-01-21 15:44:49 EET

release beta4