[
Lists Home |
Date Index |
Thread Index
]
From: John Cowan <jcowan@reutershealth.com>
Elliotte Rusty Harold scripsit:
> Optional checks worry me. This just adds another point of confusion
> like the one in XML 1.0 with non-deterministic content models, where
> some parsers accept a document and some reject it. If the check is
> optional, it should further be guaranteed to be a non-fatal error,
> even when the check is made.
That is why we say that parsers which do normalization checking must report
to the application, and emphasize that lack of normalization is not a WF
error.
But that requirement to report doesn't ensure that the application won't
stop does it? "report" doesn't seem to be specifically defined anywhere.
Also the two phrases
The option to not verify should be chosen only when the input text is
certified, as defined by [Charmod].
The option to ignore those denormalizations should not be chosen by
applications when reliability or security are critical.
Seem out of place in a normative part of the spec, they are guidelines
aimed at the end user of the application that's using the xml parser.
Such guidelines should not be in the normative prose. As is, what happens
if the user opts not to verify when the input next is _not_ certified?
Is the behaviour undefined? Partly the problem is the (5) uses of the word
"should" which appear to be being used in some colloquial sense rather
than a strict requirement in the sense of the relevant rfc's definition
of SHOULD.
David
_____________________________________________________________________
This message has been checked for all known viruses by Star Internet
delivered through the MessageLabs Virus Scanning Service. For further
information visit http://www.star.net.uk/stats.asp or alternatively call
Star Internet for details on the Virus Scanning Service.
|
