This is the mail archive of the gdb-patches@sourceware.cygnus.com mailing list for the GDB project. See the GDB home page for more information.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Bogus stepi/nexti test



> |> > The stepi/nexti tests in gdb.base/step-test.exp are completely bogus and
> |> > must be removed.
> |> 
> |> I took a different approach.  This isn't in GDB yet.  Your comments
> |> would be welcome.
> 
> Doesn't work.  The test "stepi: finish call" fails because finish does not
> work before the stack frame has been set up.

Oh --- I hadn't thought about that.  We should probably set a
breakpoint in main and continue, rather than finishing.

> While we are at it, here are a couple more portability bugs:
> 
> - call-ar-st.exp ("step inside print_all_arrays"): Prints out an
>   unterminated string, needs to ignore trailing garbage.
> - long_long.exp: Pretty much all tests are target dependent.
> - classes.exp ("print obj_with_enum (1)"): obj_with_enum is uninitialized,
>   random garbage may be printed.
>   ("print obj_with_enum (2)"): obj_with_enum.x is still unintialized,
>   random garbage may be printed.
> - miscexprs.exp ("print value of &cbig.c[0]"): Another case of an
>   unterminated string.
> - member-ptr.exp: Format of virtual tables is compiler dependent.  Casting
>   a pointer to member is compiler dependent.
> - overload.exp ("print foo_instance1"): foo_instance1.ccpfoo is
>   uninitialized, random garbage may be printed.
>   ("print foo_instance2", "print foo_instance3"): Yet another two cases of
>   an unterminated string.
> - ref-types2.exp: spelling of builtin types is compiler dependent.
> 
> It looks like portability is a foreign word at HP. :-(

Wow --- this is really helpful.  Thanks!