Re: Virtual methods table printing [Re: Expressions Status and OustandingTasks]

[Keith Seitz <keiths at redhat dot com>]

Jan Kratochvil wrote:
> Assuming one should have a simple commend to get some listing:
> (gdb) describe-me base1p
> func1: {int (derived *)} 0x40065c <derived::func1()>
> func2: {int (derived *)} 0x400672 <derived::func2()>
>
> But I am not a C++ programmer, maybe `set print object on' is sufficient?

Yeah, some way to get at RTTI would definitely be helpful for debugging 
 polymorphic designs...

Strictly speaking base1p *is* declared as "base1*". Nonetheless, I would 
be very hesitant to give access to this information via another command, 
and I prefer Jan's approach: it's simple.

If either ptype or whatis (others?) is given a pointer, then output 
actual type information for the object (in addition to its declared 
type). If given just a type name, do what gdb does today. Modify 
"print"ing of pointers as Jan suggests in his email.

I am sure there is a whole bunch of improvements to be made w.r.t. 
debugging this and other OO design paradigms that we have yet to 
discover. [One obvious one: I'd like to be able to type "list 
base1p->func1" or somesuch and have gdb list for me the correct method 
(from derived). Ditto "print" and "break".]

Keith