Re: [RFA] Improve "start" command for Ada

[Joel Brobecker <brobecker@adacore.com>]

Hello Elena,

Do you think you would have some time to look at this? The review
process stalled more than 3 months ago, and it would be nice to get
the problem fixed. In the meantime, I will KFAIL the tests in gdb.ada
that rely on the "start" command.

Thanks,

On Wed, Feb 09, 2005 at 12:05:43PM -0500, Joel Brobecker wrote:
> Hello Elena,
> 
> There is one part of the patch that hasn't been approved yet. I tried
> to explain a bit the history of the code you were concerned with, and
> I was wondering what your thoughts were. Could you have a look? (way
> at the bottom of the message - I kept the rest to keep the entire
> context together).
> 
> Thanks,
> -- 
> Joel
> 
> > On Sun, Nov 28, 2004 at 09:08:21PM -0500, Elena Zannoni wrote:
> > > Joel Brobecker writes:
> > >  > Ping? (doco already approved by Eli)
> > >  > 
> > >  > On Thu, Oct 21, 2004 at 02:09:51PM -0700, Joel Brobecker wrote:
> > >  > > 2004-10-20  Joel Brobecker  <brobecker@gnat.com>
> > >  > > 
> > >  > >         * doc/observer.texi (executable_changed): New observer.
> > >  > >         * symtab.c: Include "observer.h".
> > >  > >         (find_main_name): New function.
> > >  > >         (main_name): If name_of_main is unset, then compute it
> > >  > >         using find_main_name.
> > >  > >         (symtab_observer_executable_changed): New function.
> > >  > >         (_initialize_symtab): Attach executable_changed observer.
> > >  > >         * exec.c: Include "observer.h".
> > >  > >         (exec_file_attach): Emit executable_changed notification.
> > >  > >         * symfile.c: Include "observer.h".
> > >  > >         (reread_symbols): Send an executable_changed if appropriate.
> > >  > >         * Makefile.in (exec.o): Add dependency on observer.h.
> > >  > >         (symfile.o): Likewise.
> > >  > >         (symtab.o): Likewise.
> > >  > > 
> > >  > > Tested on x86-linux. Still fixes 1 FAIL in gdb.ada/null_record.exp.
> > >  > > 
> > > 
> > > We need a testcase where the name of the executable is changed, and
> > > this code is exercised. Otherwise ok, except for this:
> > 
> > OK, I will add something along the line of reread.exp.
> > 
> > >  > > +/* Deduce the name of the main procedure, and set NAME_OF_MAIN
> > >  > > +   accordingly.  */
> > >  > > +
> > >  > > +static void
> > >  > > +find_main_name (void)
> > >  > > +{
> > >  > > +  char *new_main_name;
> > >  > > +
> > >  > > +  /* Try to see if the main procedure is in Ada.  */
> > >  > > +  new_main_name = ada_main_name ();
> > >  > > +  if (new_main_name != NULL)
> > >  > > +    { 
> > >  > > +      set_main_name (new_main_name);
> > >  > > +      return;
> > >  > > +    }
> > >  > > +
> > >  > > +  /* The languages above didn't identify the name of the main procedure.
> > >  > > +     Fallback to "main".  */
> > >  > > +  set_main_name ("main");
> > >  > > +}
> > >  > > +
> > >  > >  char *
> > >  > >  main_name (void)
> > >  > >  {
> > >  > > -  if (name_of_main != NULL)
> > >  > > -    return name_of_main;
> > >  > > -  else
> > >  > > -    return "main";
> > >  > > +  if (name_of_main == NULL)
> > >  > > +    find_main_name ();
> > >  > > +
> > >  > > +  return name_of_main;
> > >  > >  }
> > >  > >  
> > > 
> > > Can this find_main_name become an element in the language vector? I
> > > really don't want to have a special language cases in the symtab file.
> > 
> > This has actually been discussed already. There were several messages
> > exchanged between Daniel and myself, but here are some important ones:
> > 
> >     http://sources.redhat.com/ml/gdb-patches/2004-05/msg00607.html
> >     (one potential confusion if we use the *current* language vector
> >     to determine the name of main. This is also where the suggestion
> >     of calling the Ada routine directly was suggested.
> > 
> >     http://sources.redhat.com/ml/gdb-patches/2004-05/msg00612.html
> >     (we agree that it's ok to call ada_main_name directly)
> > 
> > In short, the answer to the discussion was that this was probably the
> > best approach for now. The reason why it can't be put in the language
> > vector is that this is not a property of the language (which can vary
> > within the same program, depending on the frame), but a property of
> > the executable. None of us like this approach much, but it was something
> > that we felt sucked the least.
> > 
> > If you want, what we can do is replace the hard-coded call to
> > ada_main_name() by a loop of calls to a new language method,
> > looping on all languages until we find a positive match. That
> > way, the hard wiring to Ada disappears. But I don't think we're
> > getting much from this extra slight complexity. 
> 
> -- 
> Joel

-- 
Joel