Re: gdb.c++ testsuite 1.3: member_data_static.exp (resend)

[Daniel Jacobowitz <drow@mvista.com>]

On Wed, May 29, 2002 at 03:00:04AM -0700, Benjamin Kosnik wrote:
> 
> > As far as I can tell, gdb is reporting what's actually in the
> > executable file.  'test3' has one static member and no non-static
> > members. So perhaps gcc is broken, but I don't understand your view
> > that gdb is broken.  Can you explain more?  What do you *want* gdb to
> > print here?
> 
> Perhaps it is a gcc error, I don't know.
> 
> I'd like
> 
> mec>   (gdb) print test3
> mec>   $2 = {static data = <optimized out>}
> 
> to be
> 
> (gdb) print test3
> $2 = 
> {<gnu_obj_2> = {static test = true, static key1 = 5, static key2 = 77, 
>     static value = oriental}, _vptr.gnu_obj_2 = 0x8092efc, 
>   static value_derived = etruscian }
> 
> I have
> 
> 
> Alternatively, I'd like to be able to use explicit qualifications, to
> look at static data members, like so:
> 
> (gdb) print gnu_obj_2<long>::antiquities;
> $3 = etruscian
> 
> Seem reasonable?

Unfortunately, no.

Compile the testcase (just -c will do).  Then run `nm' over the
symbols.  Look for the `data' member; it isn't there.  <optimized out>
is correct. My C++ lawyering is a little rusty, but I believe this is
correct; it's a templated member, and not explicitly instantiated, and
not even implicitly instantiated.  I don't see a reason for it to be
emitted, and if it isn't emitted GDB can't help you figure out what its
value is.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer