Re: "error" reporting and recoverability

Tim Bray wrote:
> 
> At 02:12 PM 12/2/98 -0800, David Brownell wrote:
> >I've noticed when testing against the spec that the definition of
> >an "error" in section 1 is particularly useless when it comes to
> >defining common behaviors:
> >
> >       Error:  a violation of the rules of this specification;
> >       results are undefined.  Conforming software may detect
> >       and report an error and may recover from it.
> 
> It's my belief that this is intentional, and that the "error"
> formulation only applies in scenarios where the error either can't
> be detected deterministically; situations where things genuinely
> *are* undefined. -Tim

Hmm ... this makes me want to check the list of all "errors"
listed in the document and see if that's reall true.  Seems to
me that at least some are deterministically detectable, such as
the "content model is deterministic" case (oddly enough :-).

That sort of information would be usefully written up somewhere!

- Dave

Received on Tuesday, 8 December 1998 17:05:13 UTC