OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

dita message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Simple Types for Attributes Not in 1.3 XSDs


In response to report of difficulty constraining the attribute group for
notes in the 1.3 XSDs due to a bug in Xerces, I compared the original 1.2
XSDs to the 1.3 XSDs and realized that we (I) failed to include the simple
types used to declare some attributes in the 1.2 XSDs.

Looking at the 1.2 XSDs there are 16 simple types spread across 4 XSD
modules.

The lack of these in the 1.3 XSDs makes it difficult for users faced with
the Xerces bug to implement attribute value constraints for the elements
that use those groups (of which note is one).

Integrating this change would be a bit more involved than our errata
changes to date but it is a reliable change that should not change the
actual attribute declarations: basically replacing the literal groups in
the 1.3 XSDs with references to the corresponding simple types copied from
the 1.2 XSDs and updated as needed to reflect any changes to those
attributes from 1.2 to 1.3.

The current XSDs are, of course valid, so this is not an issue with an
actual error in the XSDs, just a difference between how they were
constructed in 1.2 and in 1.3 as a side effect of our attempts to generate
the XSDs.

The practical problem stems from a bug in Xerces whereby it fails to allow
redefinition of anonymous attributes:
https://issues.apache.org/jira/browse/XERCESJ-1670

The 1.2 DTDs do not have separate parameter entities for these attributes
(they aren't needed) and therefore the 1.3 RNGs do not have separate
patterns for these attributes (again not needed, although possibly
convenient in retrospect).

This means that the XSD generation process had nothing literal to trigger
generation of the simple types for those attributes and none of us noticed
the difference from 1.2 to 1.3. We could have implemented special case
overrides for them but I didn't realize it was necessary at the time.

I think updating the XSDs would be appropriate but it might have to wait
for the next errata cycle in order to allow sufficient time for
implementation and verification.

Cheers,

E.


--
Eliot Kimber
http://contrext.com
 







[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]