On Thursday 18 September 2008 23:51:18, Daniel Jacobowitz wrote:
> On Fri, Sep 12, 2008 at 04:38:18PM +0100, Pedro Alves wrote:
> > @@ -662,6 +662,9 @@ go32_create_inferior (char *exec_file, c
> >  #endif
> >
> >    inferior_ptid = pid_to_ptid (SOME_PID);
> > +  add_inferior (SOME_PID);
> > +  add_thread_silent (inferior_ptid);
> > +
> >    push_target (&go32_ops);
> >
> >    add_thread_silent (inferior_ptid);
>
> Is the new add_thread_silent a typo, or did you mean to remove the
> second call?

Just a typo.  Thanks.  Fixed.

> This looks otherwise OK (though I don't understand when you use
> add_inferior vs add_inferior_silent).

Hmmm, the initial thought was somewhat similar to the
use of add_thread_silent in targets that didn't add the main thread
to the list because they don't support multi-threads.  (monitor,
remote-sim).  As in, don't emit a CLI note notification if
the target pid was made up by GDB and thus should not be visible
to the user, and emit if otherwise.  This one seems to have been
missed by that logic.

But I agree that this sounds somewhat bogus.  It should be the
the target that should decide what to print...  I'll need
to do a few further adjustments for that to be possible
though.  These CLI notifications are off by default, so we
can either get rid of add_inferior_silent, or make
go32-nat.c also use it.  Any preference?  The attached does
the latter.

-- 
Pedro Alves