Hi, > Von: "Wetzler, Thomas" <[EMAIL PROTECTED]> > Datum: Tue, 2 Aug 2005 15:59:47 +0200 > > Hi there, > > I've developed a schema-definition with the tool "XML-Spy" on my > workstation. It works fine without any failures:
W3C XML Schema is challenging to implement - XML-Spy might have a flaw here. If you want to be sure that your schema is valid, then try validating it with various processors like Xerces, XSV, MSXML, etc. If they report different results - and they do often - then consult the schema mailing lists, or your mighty company's schema consultant - Siemens should have one ;-) [...] > <xs:complexType name="type_subelement"> > <xs:choice> > <xs:sequence> > <xs:element name="header" type="columns" > fixed="AN"/> > <xs:element name="content" > type="type_an"/> > </xs:sequence> > <xs:sequence> > <xs:element name="header" type="columns" > fixed="PN"/> > <xs:element name="content" > type="type_pn"/> > </xs:sequence> > </xs:choice> > </xs:complexType> [...] > Warning: complex type 'type_subelement': The content model is > not determinist. in > /wir/searchservertest/src/load_db/xml_parser/FtcinCreator.php on line 37 > > Warning: Invalid Schema in > /wir/searchservertest/src/load_db/xml_parser/FtcinCreator.php on line 37 > Validate FAILED > > The library seems to have difficulties with the <xs:choice><xs:sequence> > combination. XML-Spy has no problem with that. > > Does anyone have an idear? Your schema seems to be invalid. Xerces 2.7.1 reports: groups-nested-1.xsd:7,48: (Error) cos-element-consistent: Error for type 'type_subelement'. Multiple elements with name 'content', with different types, appear in the model group. groups-nested-1.xsd:7,48: (Error) cos-nonambig: header and header (or elements from their substitution group) violate "Unique Particle Attribution". During validation against this schema, ambiguity would be created for those two particles. Although XSV does not bark at it, I would trust the results of Xerces here, since, AFAIK, XSV already tries to implement a way of unique particle attribution (UPA), which would probably fit for the next version of the XML Schema spec. It seems that you want to implement something like (pseudo code): if (header == "AN") then content.type = type_an else if (header == "PN") then content.type = type_pn This is not possible with XML Schema, which is in turn revealed by XSV with the following example instance: <element xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; xsi:noNamespaceSchemaLocation="groups-nested-1.xsd"> <subelement> <header>PN</header> <content>1000</content> </subelement> </element> XSV 2.10-1 reports: groups-nested-1.xml:5,3: fixed value did not match: PN!=AN A <choice> in XML Schema is usable at the element QName level only, i.e. processors will not try other "choices" if the node's content or attributes are invalid, but just reject the instance. Have a look at RelaxNG or Schematron. I don't know both of them much; maybe they provide the mechanisms you need. Or, once again, consult your mighty company's schema consultant. Cheers, Kasimier _______________________________________________ xml mailing list, project page http://xmlsoft.org/ [email protected] http://mail.gnome.org/mailman/listinfo/xml
