Re: Local Context [was: Why I moved from Forbid to Literal]

I agree absolutely with your conclusion:

>  I prefer the literal approach because it is just that, and the easiest to
> avoid screwing up. Using a context is the most 'logical'
> approach

Perhaps that much agreement is enough. Certainly after many very long days we
ought perhaps to accept peace where we find it. Nevertheless we will not be able
to avoid picking over your parting shot

> assuming you all agree with my definition of the term 'context'

however likely that is to get us back into cyclic wrangling. FWIW, I am happy to
let the 'context' define itself:  to me, it is effectively the sum of all those
things which influence the processing at the local node. That context is simply
not the same when the local node is performing an XSLT transform as it is when
that node is running some unique idiosyncratic bit of software--some custom
application on top of XML-family modules. When we see production implementations
of XML schema processing, query processing, link processing, etc. it is likely
that, even without idiosyncratic custom applications running and even where
differing processing of only a single instance document is compared, salient
constituents of the local context will differ even between local processes
performed only by XML-family modules:  we have seen examples of such differences
already. There will also be differences in locally elaborated semantics--and
therefore locally-executed processing--between instances of documents which have
the same Infoset except for trivial, literal syntactic differences such as the
alphabetic case differences of namespace names which we have discussed. Those
are the inevitable consequences of the 'literal' position in this debate, and
consequences which I am happy to accept. The question is whether you--and
others--will accept those consequences as necessary to an agreement on a
definition of the term 'context'.

Respectfully,

Walter Perry

Received on Friday, 30 June 2000 19:18:59 UTC