On 1/2/25 12:52, Alessandro Vesely wrote:
> On Thu 02/Jan/2025 00:15:19 +0100 Daniel K. wrote:
>>   <xs:complexType name="ReportMetadataType">
>> - <xs:all>
>> +  <xs:sequence>
> 
> That change was by Matt Wander in 2021 on gdoc
> https://docs.google.com/document/d/1Fh18iswRu4WJxnN6LBPIpVBncwhfJODOQxoeYH__Z-Q/
> 
> It looked like a harmless simplification, given the nature of those fields.

Thanks, I somehow bungled the arguments to interdiff, so I quoted the
reverse of the actual change that was from "sequence" => "all".

I agree with the rationale given by Matt.

"error" was later put back, but with maxOccurs="1" which is compatible
with "all".


>>   <xs:complexType name="RowType">
>> - <xs:sequence>
>> +  <xs:all>
> 
> This has always been unordered, AFAIK.

Yes, but it is now "sequence".

The same argument holds for RowType (no unbounded elements), and I will
add a commit changing it back to "all", the way ut was in RFC7489.


> The only reason I can think of is to ease poor parsers.  As they've always 
> been <xs:all>, should we check if report generators actually follow the order 
> given in the schema?

As long as we loosen the restrictions, I don't think it matters.


Daniel K.

_______________________________________________
dmarc mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to