search

w3c / xsdtests

W3C XML Schema 1.1 test suite
17 stars 7 forks source link

Attribute notQName allowed for the xs:any of an openContent/defaultOpenContent #1

Closed brunato closed 5 years ago

brunato commented 5 years ago

Hello, In the XSD 1.1 tests the particle xs:any inside an openContent or defaultOpenContent is considered to allow the notQName attribute, so for example the cases xsdtests/saxonData/Wild/wild074.xsd and xsdtests/ibmData/instance_invalid/S3_4_1/s3_4_1ii04.xsd are signed as valid schemas.

But in the XSD 1.1 meta-schema the particle xs:any declare for an openContent/defaultOpenContent does not allow the attribute notQName. So which is wrong?

Thank you

michaelhkay commented 5 years ago

You appear to be correct that the schema documents used in these tests do not conform to the published S4SD.

I can't see any technical reason for disallowing the notQName attribute on any xs:any wildcard that appears within openContent. One might argue that it serves no useful purpose, but the XSD working group generally avoided "paternalism" - that is, putting an artificial restriction in the spec merely because they didn't think something was useful. So it could be an oversight in the S4SD. We will never know.

Michael Kay

On 8 Aug 2019, at 10:35, Davide Brunato notifications@github.com wrote:

Hello, In the XSD 1.1 tests the particle xs:any inside an openContent or defaultOpenContent is considered to allow the notQName attribute, so for example the cases xsdtests/saxonData/Wild/wild074.xsd and xsdtests/ibmData/instance_invalid/S3_4_1/s3_4_1ii04.xsd are signed as valid schemas.

But in the XSD 1.1 meta-schema https://www.w3.org/TR/xmlschema11-1/#normative-schemaSchema the particle xs:any declare for an openContent/defaultOpenContent does not allow the attribute notQName. So which is wrong?

Thank you

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/w3c/xsdtests/issues/1?email_source=notifications&email_token=AASIQIWIAAMFVW67DBA7UFLQDPSGBA5CNFSM4IKIJ3GKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HEDEZSA, or mute the thread https://github.com/notifications/unsubscribe-auth/AASIQIRT2554IRILNA3PM4LQDPSGBANCNFSM4IKIJ3GA.

brunato commented 5 years ago

Yes, it should be an oversight due to different syntax that notQName has in xs:any and in xs:anyAttribute. In this case, for using the meta-schema in validation may have sense to redefine the openContent to include also the notQName in the particle.

Thanks