Hi,

$SUBJECT concerns an assertion failure that is triggered when using 
pretty-printing with varobj. The cause of the problem is actually pretty 
simple: cplus_describe_child knows nothing about pretty printing.

As a result, it strictly enforces that the immediate children of a root 
varobj whose type is a class/struct must be one of the "fake" children. 
When asking for an index greater than 2 (CPLUS_FAKE_CHILDren are indices 
0, 1, 2), the code asserts because of an unknown fake child.

The solution is to teach cplus_describe_child how to fetch the N'th 
child from the pretty-printed varobj.

No regressions on x86_64 native and native-gdbserver.

Keith

ChangeLog
2013-07-08  Keith Seitz  <keiths@redhat.com>

	PR varobj/15166
	* varobj.c (get_pretty_printed_element_index): New function.
	(cplus_describe_child): Add handling for pretty-printed varobjs.

testsuite/ChangeLog
2013-07-08  Keith Seitz  <keiths@redhat.com>

	PR varobj/15166
	* lib/mi-support.exp (mi_list_varobj_children_range): Ignore
	a display hint in the expected output.
	* gdb.python/py-pp-varobj-update.cc: New file.
	* gdb.python/py-pp.varobj-update.exp: New file.
	* gdb.python/py-pp-varobj-update.py: New file.