Bug 14618

Hi Alex,

openteacher runs fine here without a single warning on the console, on Mageia Linux 5 x86-64 Cauldron (Are you running it on i586?), on top of KDE 4. Can you check whether:

1. It happens in a new UNIX user account.

2. It happens with an English locale.

Regards,

-- Shlomi Fish

> 1. It happens in a new UNIX user account.

Yes

> 2. It happens with an English locale.

Yes

(In reply to Alex Loginov from comment #2)
> > 1. It happens in a new UNIX user account.
> 
> Yes
> 
> > 2. It happens with an English locale.
> 
> Yes

I now tried running openteacher inside a fresh i586 Mageia 5/Cauldron VM , and it ran fine:

http://www.shlomifish.org/Files/files/images/openteacher-in-cauldron-i586.png

So I have no idea how to reproduce your problem. Maybe you have corrupt files.

Regards,

-- Shlomi Fish

Better output if to install faulthandler:

urpmi python-pip
pip install faulthandler
[user@localhost ~]$ openteacher
openjdk version "1.8.0_40"
OpenJDK Runtime Environment (build 1.8.0_40-b12)
OpenJDK Server VM (build 25.40-b16, mixed mode)

(python:18352): GLib-GObject-WARNING **: cannot register existing type 'GdkDisplayManager'

(python:18352): GLib-CRITICAL **: g_once_init_leave: assertion 'result != 0' failed

(python:18352): GLib-GObject-CRITICAL **: g_object_new: assertion 'G_TYPE_IS_OBJECT (object_type)' failed

(python:18352): GLib-GObject-WARNING **: invalid (NULL) pointer instance

(python:18352): GLib-GObject-CRITICAL **: g_signal_connect_data: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed

(python:18352): GLib-GObject-WARNING **: invalid (NULL) pointer instance

(python:18352): GLib-GObject-CRITICAL **: g_signal_connect_data: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed

(python:18352): GLib-GObject-WARNING **: cannot register existing type 'GdkDisplay'

(python:18352): GLib-CRITICAL **: g_once_init_leave: assertion 'result != 0' failed

(python:18352): GLib-GObject-CRITICAL **: g_type_register_static: assertion 'parent_type > 0' failed

(python:18352): GLib-CRITICAL **: g_once_init_leave: assertion 'result != 0' failed

(python:18352): GLib-GObject-CRITICAL **: g_object_new: assertion 'G_TYPE_IS_OBJECT (object_type)' failed
Fatal Python error: Segmentation fault

Current thread 0xb7335940 (most recent call first):
  File "/usr/lib/python2.7/site-packages/openteacher/modules/org/openteacher/gui/gui.py", line 433 in run
  File "/usr/lib/python2.7/site-packages/openteacher/modules/org/openteacher/uiController/uiController.py", line 145 in run
  File "/usr/lib/python2.7/site-packages/openteacher/modules/org/openteacher/event/event.py", line 35 in send
  File "/usr/lib/python2.7/site-packages/openteacher/modules/org/openteacher/execute/execute.py", line 143 in execute
  File "/usr/lib/python2.7/site-packages/openteacher/openteacher.py", line 48 in run
  File "/bin/openteacher", line 27 in <module>
Segmentation fault

Shlomi, hi,

You're lazy :) I contacted with author yesterday and he reproduced bug and I have solution today.

Answer from author:

I managed to track down the segfault. The following is a workaround:
- su
- mv
/usr/lib/python2.7/site-packages/openteacher/modules/org/openteacher/hiddenBrowser
/usr/lib/python2.7/site-packages/openteacher/modules/org/openteacher/_hiddenBrowser

This disables a feature that was never meant as much more than a simple
easter egg, users are not likely to miss it. There's already a bug
report open that would mitigate the problem a bit on OT's side. [1] The
problem ultimately is in QtWebKit/PyQt4, though, not in OpenTeacher.
This small but valid python script proves that: [2]. I recommend using
that code to open a bug report against QtWebKit.

[1]: https://bugs.launchpad.net/openteacher/+bug/1206600
[2]: http://paste.ubuntu.com/9313523/

Thanks again for reporting and I hope this helps,

-- 
Marten de Vries

Hello Alex,

I am still unable to reproduce your bug and the script in the paste runs fine from what I see: http://paste.ubuntu.com/9313523/ . Do you know why I cannot reproduce it? I did everything I can think of. Is some web site blocked or filtered where you are?

Regards,

-- Shlomi Fish

Yesterday I corresponded with author: many reasons why this bug is not always reproducible: KDE, phonon, tesseract, cuneiform, etc. So, your preinstalled pkgs in distro are not the same with mine when I have a lot of preinstalled pkgs 3500+.

> Is some web site blocked or filtered where you are?
Yes. But I don't think it's a reason for segfault. Author reproduced bug in the Netherlands, where everything is OK with the Internet.

Will you apply workaround in openteacher pkg?

(In reply to Alex Loginov from comment #7)
> Yesterday I corresponded with author: many reasons why this bug is not
> always reproducible: KDE, phonon, tesseract, cuneiform, etc. So, your
> preinstalled pkgs in distro are not the same with mine when I have a lot of
> preinstalled pkgs 3500+.
> 
> > Is some web site blocked or filtered where you are?
> Yes. But I don't think it's a reason for segfault. Author reproduced bug in
> the Netherlands, where everything is OK with the Internet.
> 
> Will you apply workaround in openteacher pkg?

Done:

http://pkgsubmit.mageia.org/

Fixed in openteach-3.2-6.mga5 . RESOLVing as FIXED.

Thanks for the report.

Regards,

-- Shlomi Fish

Resolution: (none) => FIXED
Status: ASSIGNED => RESOLVED

Shlomi, It seems I found reason - flash-player-plugin.
A version of Qt doesn't mix with a version of Flash.
https://bugs.webkit.org/show_bug.cgi?id=90065
So, I reopened bug. Will you reproduce now after urpmi flash-player-plugin?

Anyway we needs workaround as enabled for openteacher while we will not find other better fix.

Status: RESOLVED => REOPENED
Resolution: FIXED => (none)
CC: (none) => anssi.hannula

(In reply to Alex Loginov from comment #9)
> Shlomi, It seems I found reason - flash-player-plugin.
> A version of Qt doesn't mix with a version of Flash.
> https://bugs.webkit.org/show_bug.cgi?id=90065
> So, I reopened bug. Will you reproduce now after urpmi flash-player-plugin?
> 

I can reproduce the problem after "urpmi flash-player-plugin". Good job on the investigation. 

> Anyway we needs workaround as enabled for openteacher while we will not find
> other better fix.

The workaround is already in the package.

(In reply to Alex Loginov from comment #9)
> Shlomi, It seems I found reason - flash-player-plugin.
> A version of Qt doesn't mix with a version of Flash.
> https://bugs.webkit.org/show_bug.cgi?id=90065
> So, I reopened bug. Will you reproduce now after urpmi flash-player-plugin?
> 
> Anyway we needs workaround as enabled for openteacher while we will not find
> other better fix.

should we assign this bug to a different package?

I added to CC flash-player-plugin maintainer. Feel free to add somebody yet, you know Mageia people better, than I. It's total problem for distro and will be reproduced in other pkgs, so additional attention is good.

Assigning to the flash-player-plugin package then.

Source RPM: openteacher => flash-player-plugin

Reassigning to the flash-player-plugin maintainer.

Assignee: shlomif => anssi.hannula

Adding KDE team in CC since it's apparently a problem between flash and Qt.

CC: (none) => lmenut, mageia
Whiteboard: (none) => MGA5TOO
Summary: openteacher crashing => openteacher crashing (flash-player-plugin vs Qt issue)

I guess this bug is so old we can close it?
No further action for four years.
And soon flash is no more.

CC: (none) => fri

Flashplayer is EOL since 31.12.2020 and got permanently blocked from execution 12.01.2021.

flash-player-plugin is not available anymore in MGA8 onwards.

Status: REOPENED => RESOLVED
Resolution: (none) => OLD