Re: Error reporting should be improved

[Peter Rosin]

Den 2009-12-30 00:41 skrev Bob Friesenhahn:
> On Tue, 29 Dec 2009, Peter Rosin wrote:
> > I'm just saying that it is not going to be easy to deduce if an
> > error is allowed to be overwritten.
> >
> > Have a look at libltdl/loaders/dlopen.c:vm_open(). It basically
> > calls dlopen and sets the error to CANNOT_OPEN if dlopen fails.
> > The ways I see to dig out the underlying failure is to either call
> > dlerror and parse the string (Eeek!) or replicate the dlopen
> > internals and arrive at the same cause as dlopen did (EEEK!).
>
> If dlerror() is available it is supposed to be ignoring the supplied 
> CANNOT_OPEN code and invoking dlerror() instead:
>
> #if defined(HAVE_DLERROR)
> #  define DLERROR(arg)  dlerror ()
> #else
> #  define DLERROR(arg)  LT__STRERROR (arg)
> #endif
>
> #define DL__SETERROR(errorcode) \
>         LT__SETERRORSTR (DLERROR (errorcode))

Yes, I saw that later, but it doesn't change much. Any scheme has
to deduce if an error can be overridden without help from the
loaders. I now think you were correct in your first message where
you said that we need a flag or a full blown error stack...

Cheers,
Peter

--
They are in the crowd with the answer before the question.
> Why do you dislike Jeopardy?