If you have a stack of identified items (or a stack of failed-to-identify items), and a stack of equivalent unidentified items, and use the click-examine to identify them (I forget the exact command the GTK client sends on click-examines, but that's what I was using), if the stack being examined is merged into another stack, you get an examine description with a first line of "That is: (null)".
It looks like it might be due to where in the identification process we try to merge objects. The attempt to describe the pile of objects is still pointing to the invalid freed object from before the pile merge. There does not appear to be a simple way to propagate a pointer change out of identify_object_with_skill() without some non-negligible refactoring.