Comment # 7 on bug 981011 from Michael Andres

(In reply to Dominique Leuenberger from comment #5)
> That sseems PK passing the objects from libzypp over an iterrator - as there
> was no change just recently in PK I have to assume a change in libzypp might
> cause this.
> 
> @Michael: can you recall a change that might cause this?

No.


But if I look at the PK backend code and the BT location is correct:

> #5  backend_get_update_detail_thread (...) at pk-backend-zypp.cpp:2454

> 2454: make<ResObject>(solvable)->description().c_str (),	// update-text

The surrounding for-loop creates 'sat::Solvable solvable' from package_ids
which seem to have been determined and stored before:
> 2406    sat::Solvable solvable = zypp_get_package_by_id (package_ids[i]);
> 2407    MIL << package_ids[i] << " " << solvable << endl;

There's no check whether the sat::Solvable returned by get_package_by_id is
actually valid:
> if ( ! solvable ) 
> { 
>    // Previously stored package_id no longer matches any solvable.
>    // May happen e.g. if repos meanwhile changed (add/remove/refresh)
> }

In this case ' make<ResObject>(solvable)' in line 2454 is a nullptr.




If the backend logging is on, the log line written @2407 (MIL <<...) should
tell whether this is what happened.