Re: tests

On Wed, 4 Aug 2004 07:16:21 -0700, David Brownell <[email protected]> wrote:
> On Tuesday 03 August 2004 20:15, Musachy Barroso wrote:
> > Some of this SAX tests are driving me crazy..like this one, on this input:
> >
> > <?xml version='1.0' standalone='yes'?>
> >     <!DOCTYPE attributes SYSTEM "../valid/sa.dtd" [
> >         <!--
> >           attribute needs defaulting
> 
> .... and since it needs defaulting, clearly "standalone = yes" is wrong
> and the document is invalid.
> 
> >          -->
> >     ]>
> > <attributes/>
> > <?pi equals three?>
> >
> 
> In this case, since it needed to default to <attributes token='a'/>,
> clearly omitting the DTD would change the output and so the
> document would no longer be standalone:  parser output would
> NOT BE THE SAME when omitting the DTD's external subset.
> 
> 
> > the SAX test is expecting the parser to accept the xml file(!?), the
> > header of the test reads: " Tests the Standalone Document Declaration
> > VC, ensuring that attributes needing defaulting cause a validity
> > error."
> 
> Accept it, sure.  Accept it as valid, no ... it's invalid according to the
> original intent of the XML specification.  Were you validating?
> Reading the external subset of that DTD?

yeah...accept it as valid I meant.

> The problem is that the "standalone" declaration was always very
> poorly defined ... and the behavior of attribute defaulting with respect
> to that declaration was _very badly_ defined.
> 
> And to top it off, the first several errata to that part of the XML spec just
> added more ambiguity there.  The W3C folk first tasked with resolving
> such things were dodging that issue pretty seriously, when they weren't
> ignoring it ... I remember it took over a year to get them to even respond
> to that one, at which point their alleged "response" was off topic.
> 
> I got the impression that vendor politics were behind a lot of that
> early mess, though that over-a-year delay was maybe more because
> they didn't want to revise XML 1.0 that soon after releasing it.  And the
> XML conformance test process sure acted for a while like it was being
> used as a vehicle to "bless Xerces" rather than actually resolve the
> spec bugs that turned up ... not just those for "standalone".
> 
> - Dave
> 
I just wrote to them  'cause there are actually 10 tests in this
condition (sun/invalid/not-sa...).

thanks

musachy


-- 
"Hey you! Would you help me to carry the stone?"
                                                         Pink Floyd