[
Lists Home |
Date Index |
Thread Index
]
Hi,
On Thu, 2005-11-24 at 11:00 +0000, Paul Spencer wrote:
> You are correct in saying that XML Spy 2004 does not detect this error, but
> 2005 does. This (and inadequate testing) has led to at least one set of
> invalid schemas being published as a "standard".
>
> When you use redefine, the new definitions must be legitimately derived from
> the types you are redefining. Your simple type is derived from an xs:string.
> You just need to change this to derive it from the original type.
>
> <xs:redefine schemaLocation="Events_1.6.xsd">
> <xs:simpleType name="RecurrenceIDRangeType">
> <xs:restriction base="RecurrenceIDRangeType">
> <xs:enumeration value="ThisAndFuture"/>
> <xs:enumeration value="ThisAndPrior"/>
> </xs:restriction>
> </xs:simpleType>
>
> Your complex type is not a valid restriction, and I don't think you can
> create one to do what you want.
True, the complex type restriction won't work.
If XML Schema's unique particle attribution (UPA) constraints [1] were
not built to reflect a _deterministic_ finite state automaton, then you
could define your base type as follows:
<xs:element name="a"/>
<xs:element name="b"/>
(warning: this <xs:choice> is _not_ _valid_)
<xs:choice minOccurs="0">
<xs:element ref="a"/>
<xs:element ref="b"/>
<xs:sequence>
<xs:element ref="a"/>
<xs:element ref="b"/>
</xs:sequence>
</xs:choice>
... removing the empty set using a restriction would only need the
minOccurs to be set to 1:
(warning: this <xs:choice> is _not_ _valid_)
<xs:choice>
<xs:element ref="a"/>
<xs:element ref="b"/>
<xs:sequence>
<xs:element ref="a"/>
<xs:element ref="b"/>
</xs:sequence>
</xs:choice>
This would be a valid _restriction_. But, as already mentioned, both
models violate the UPA constraints in the way that the automata,
reflecting those models, would be non-deterministic.
[1] http://www.w3.org/TR/xmlschema-1/#cos-nonambig
---
Potential bug in Saxon 8.5.1: I noticed that it does not fire an
UPA error for the above scenario; on the other hand it does fire the
error if the @ref attributes are substituted for @name attributes.
A "different/same element declaration" issue?
---
[...]
Regards,
Kasimier
|
