Daniel Jacobowitz wrote:

> On Wed, Nov 29, 2006 at 05:41:13PM +0300, Vladimir Prus wrote:
>> +# Creates varobj named NAME for EXPRESSION.
>> +# Name cannot be "-".
>> +proc mi_create_varobj { name expression } {
> 
>> +# Updates varobj named NAME and checks that all varobjs in EXPECTED
>> +# are reported as updated, and no other varobj is updated.
>> +# Assumes that no varobj is out of scope and that no varobj changes
>> +# types.
>> +proc mi_varobj_update { name expected } {
> 
>> +proc mi_check_varobj_value { name value } {
> 
> I would recommend giving these an argument for the test name, as some
> of the other helper functions do.  Otherwise, if you call them twice in
> a row with the same arguments (which your test does) then you've got
> two tests with the same name.
> 
> On Thu, Nov 30, 2006 at 10:31:07AM +0300, Vladimir Prus wrote:
>> >         * gdb.mi/mi-cpp.cpp: New file.
>> >
>> > gdb.mi/mi-var-cp.exp?
>> > gdb.mi/mi-var-cp.cc?
>> >
>> > Its for variable objects, and for consistency.  The testsuite has the
>> > directory gdb.cp and it's populated with *.cc files
>> 
>> I've no problems with 'cp' and '.cc'. I don't think that '-var-' is good
>> -- now, there is just single MI test dealing with C++, so I want a
>> testcase that will accumulate all C++ specific things, not necessary
>> related to variable objects. But I don't care much.
> 
> I'd rather keep varobj tests easily identifiable; please use Nick's
> suggestions.

Here's revised patch. Ok for HEAD?

- Volodya

        * varobj.c (varobj_create): Don't call release_value.
        (varobj_set_value): Likewise.
        (install_new_value): Call coerce_ref and release_value
        on the value. Add asserts.

        testsuite/
        * lib/mi-support.exp
        (mi_runto): Accept "()" after function name.
        (mi_create_varobj): New function.
        (mi_varobj_update): New function.
        (mi_Check_varobj_value): New function.
        * gdb.mi/mi-var-cp.exp: New file.
        * gdb.mi/mi-var-cp.cc: New file.