Re: Re: Error reporting capabilities for libxc

["Daniel P. Berrange" <berrange@redhat.com>]

On Mon, Oct 23, 2006 at 11:15:21PM +0100, Ian Pratt wrote:
> > > Would we be better off returning an error code and a set of
> parameters,
> > > requiring a call-back into the library to get the string?
> > 
> > That assumes that there is a static mapping between an error code
> > and the error description, which there isn't. The error description
> > can contain info about actual bits of metadata which were incorrect.
> > For example when reporting an invalid ELF architecture, it can tell
> > you exactly what ELF arch was found & what was expected.
> 
> That's precisely my point. The various bits of metadata can be
> parameters passed back along with the error code. It's possible that
> some callers will be able to do things with the metadata that are more
> useful than just printing the string. I guess the error parameters could
> be returned as a va_list. Many callers would just call back into the
> library to get the appropriate error string.

Ahh, I mis-read your comment a little. That is certainly one possibility
although I'm not entirely a fan of va_list as an API - perhaps either a
struct with a handful of generic data fields for returning metadata, or
a union switched on the error code would work - it'd be nice to be able
to copy all the bits around without needing memory allocation in any of
the error handling paths.

Regards,
Dan.
-- 
|=- Red Hat, Engineering, Emerging Technologies, Boston.  +1 978 392 2496 -=|
|=-           Perl modules: http://search.cpan.org/~danberr/              -=|
|=-               Projects: http://freshmeat.net/~danielpb/               -=|
|=-  GnuPG: 7D3B9505   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505  -=|