Bug 1115589

I would very much like to see the upstream solution for this problem, it seems one may come up in bug [1]. Since this is not a critical issue, I'm moving this bug to 7.3.

[1] http://bugs.python.org/issue12902

No.
We can help with workarounds in specific cases where this poses a problem, but changing what help('modules') does is not realistic in Python 2.

** We can do this _if_ we have proper DIY way for users to find out which all modules are in conflict and based on it, how to workaround.

Current status:

* The original bug is fixed, so the following should work:
    # yum group install 'Server with GUI'
    # python -c "help('modules')"

* bug 1451299 was filed for pygtk2 regarding interaction between gtk3 and gtk2 libraries, i.e. the fact that the following will still cause a segmentation fault:

     python -c 'from gi.repository import Gtk; from gtk import _gtk'

* Importing gtk will fail with a Python traceback when a graphical display is not available. So, the following will fail on a typical server installation:

    python -c "import gtk; help('modules')"

  This is expected. The error happens in the first statement; `python -c "help('modules')"` will still work.
  The "import gtk" statement will also fail when pygtk2 is not installed.

* Other modules (or combinations of modules) can still make help('modules') fail, since pydoc imports all available modules, which executes their code. Unfortunately, we can only address these issues individually, if they are reported.

I've just tested it on RHEL 7.4 and indeed the reproducer no longer works:

    # yum group install 'Server with GUI'
    # python -c "help('modules')"

I verified that 'orca' is installed as well and still the crash is not reproduced.

Is the current status/workarounds enough for the customer?

We still have customers reporting this, request to recheck issue.

The modules command is running under python interface,
but still we are getting those messages :

/usr/lib64/python2.7/site-packages/gobject/constants.py:24: Warning: g_boxed_type_register_static: assertion 'g_type_from_name (name) == 0' failed
  import gobject._gobject

The failure is a result of interaction between several modules.

What other Python modules does this customer have installed? Can you post the output of:

    ls /usr/lib64/python2.7/site-packages/

Also, what command is being run? Is it `python -c "help('modules')"`, or something else?

Also, let me re-open this bug. Closed bugs don't show up on trackers, so we're quite likely to miss any further comments on them.

Thanks for revisiting this.

I am running just python -c "help('modules')"

interestingly this time coredump as well.

python -c "help('modules')"

Please wait a moment while I gather a list of all available modules...

/usr/lib64/python2.7/site-packages/gobject/constants.py:24: Warning: g_boxed_type_register_static: assertion 'g_type_from_name (name) == 0' failed
  import gobject._gobject
/usr/lib64/python2.7/site-packages/gtk-2.0/gtk/__init__.py:40: Warning: specified class size for type 'PyGtkGenericCellRenderer' is smaller than the parent type's 'GtkCellRenderer' class size
  from gtk import _gtk
/usr/lib64/python2.7/site-packages/gtk-2.0/gtk/__init__.py:40: Warning: g_type_get_qdata: assertion 'node != NULL' failed
  from gtk import _gtk
Segmentation fault (core dumped)

python.x86_64 0:2.7.5-58.el7 python-libs.x86_64 0:2.7.5-58.el7 

updated from  python-libs-2.7.5-48.el7.x86_64 python-libs-2.7.5-48.el7.x86_64 

attaching ls output as ls.out

Created attachment 1342221 [details]
ls

The conflicting modules are `gi` and `gio`. I've filed bug #1506164 on pygobject2.

This particular problem is now tracked in bug 1506164. Feel free to re-open if there are other similar issues.

Two other issues found:
pykde4 and scipy interaction - bug 1533516
pykde4, python-libteam and python-ethtool interaction - bug 1533517

It might actually be pykde4 problem, but I don't want to jump to conclusions.

In comment to first issue maintainer of scipy package points out to upstream issue that has not been discussed here: https://bugs.python.org/issue11995

Any update here regarding the comment #28? Reading bug #1533516, it still doesn't seems to me like anything wrong is done in python component, so can we close this bug?

Right, we can close this bug -- crashes come from (combinations of) other components.