Re: Heuristics

At 2002-07-11 20:40, Martin Duerst wrote:
>Hello Michael,
>
>Many thanks for your comment. We have received similar comments,
>and have already decided to make some changes, although we don't
>yet know the exact wording.
>
>We think that the Appendix of XML 1.0 does not give heuristics,
>in that no two implementations will interpret a document as
>being encoded in different encodings. (there may of course
>be differences in what encodings the two implementations accept).
>
>By heuristics, we mainly meant statistically or otherwise
>judgement-based or guessing-like procedures that won't be
>implemented consistently in different implementations.
>Do you have any idea of how to express this in a better way?

A definition of "heuristic" specifying that by "heuristic" the
spec means procedures which have wholly or partly indeterminate
results, in particular procedures based on statistical sampling
of the input.  I think providing a definition would allow you
to make the point clearer without burdening the prose of the
relevant rule.

I'm glad to know that Appendix F is safe.

Michael

Received on Saturday, 13 July 2002 16:55:14 UTC