SV: SV: SV: [ubl] Revised discussion paper on UBL 2.0 subsets, extensions, versions, validation and interchange

["Bryan Rasmussen" <BRS@itst.dk>]

Hi Ken

>I am willing to accept your assessment and can 
>withdraw the proposal based on 
>non-implementability (is that a word?) ... can 
>anyone else offer a comment on the 
>non-determinism?  Does anyone have an opinion 
>regarding which processor is correct?

My opinion is basically that XSV is correct on this matter, in that I would
tend to trust Henry Thompson in matters of non-determinism, problems with XSV
has generally not been on this stuff and it seems to me that this is one of
the things the whole FSA based approach is ideally suited for. 


Thinking about it abstractly:

It is non-deterministic because we cannot be sure if we are supposed to do
strict or skip. 

However if it validates by strict then the non-determinism does not apply
(one could argue this, but it feels really weird). In other words, the schema
is non-deterministic but we can be sure that the instance is correct anyway
because if it is correct by strict it is surely correct by skip as well. I
think that explains the otherwise strange output of XSV. However it has been
my experience that most processors would take this as non-deterministic and
wrong, and IIRC XSV at one point did take it as both non-deterministic and
wrong. For all these reasons I would like to avoid it, and while it might be
theoretically implementable in that way I think in most real world
applications it will be non-implementable. 


Cheers,
Bryan Rasmussen