Description of problem: system-config-authentication seg faults at command line. Version-Release number of selected component (if applicable): authconfig-gtk-5.3.8-1 How reproducible: Always Steps to Reproduce: 1. Login as root to a virtual terminal 2. Run 'system-config-authentication' Actual results: error, and segfault Expected results: Should work, or explain why not (really, it should work, in case authentication is sad and not letting one log into X to run it there). Additional info:
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux release. Product Management has requested further review of this request by Red Hat Engineering. This request is not yet committed for inclusion in release.
Bug seems to be deeper than just flat-out brokenness. Works for me on a i386 desktop (Suzanne's machine is a i386 laptop).
Created attachment 137427 [details] errors that go to the screen when s-c-authenticaion is run
The problem is coming from elsewhere than system-config-authentication. Are you running it from a linux text console? What echo $DISPLAY says there? The system-config-authentication (alias authconfig-gtk) tries to import gtk module which should properly error out and not segfault when $DISPLAY is not set. The error is then catched by s-c-a and authconfig is execed instead of it. Workaround - use 'authconfig' command instead of 'system-config-authentication'.
It appears s-c-a expects an exception to be raised when importing gtk without a display. That does not seem to happen. Even if it would happen though, s-c-d tries to run authconfig, when it should probably run authconfig-tui ?
It raised an exception at least in FC5. It must run authconfig because authconfig-tui is deprecated and is not being enhanced with new options anymore (smart card support for example). What is the proper way to detect no X access with current pygtk?
It seems system-config-display has similar problems
This is due to a (completely broken IMHO) change in upstream pygtk. See http://bugzilla.gnome.org/bugzilla/show_bug.cgi?id=316877 for the details and my attempt to talk sense into them :( I think that given where we are, it probably makes sense to revert the behavior in our packages for FC6 and then begin the painful process of changing apps for post-FC6.
There's and extra /bugzilla/ in your url, Jeremy. The upstream bug links to http://live.gnome.org/PyGTK/WhatsNew210 which describes how to catch "DISPLAY not set" with 2.10
Bah, that's what I get for just typing and not actually checking the URL :-) And yes, I know they have an all new way of doing it. But the whole "break existing apps" is completely and utterly broken. And I really don't see how we can manage to _successfully_ go through and fix all of the apps we ship in the time we've got.
yes, you are right, going back to throwing an exception seems the only way to fix system-config-* in time for fc6
*** Bug 208911 has been marked as a duplicate of this bug. ***
Fixed in pygtk-2.10.1-4