This issue is not in scope of the PRD02 public review.

However, Alex at the glance seems to be right: SMIL states in 10.3.1:

> For the purpose of parsing the endsync argument value, first, last, all, and media are reserved words and must be escaped with
> a backslash in order to be used as Id-value's.

Such escaped values are not permitted by the ODF schema. SMIL defines a normative DTD. Here, the endsync attribute has the type "CDATA", which means that there are no restrictions on the values at all, but also no ID/IDREF semantics.

But if we look at the informative SMIL XSD schema, then "endsync" is defined there as:

<!-- utility type for endsync attribute arg values -->
<simpleType name="endsyncArgType">
<restriction base="string">
<enumeration value="first"/>
<enumeration value="last"/>
<enumeration value="all"/>
<enumeration value="media"/>
</restriction>
</simpleType>

<!-- define the endsyncTimingAttrsType -->
<simpleType name="endsyncTimingAttrsType">
<union memberTypes="smil20:endsyncArgType IDREF"/>
</simpleType>

<!-- define the endsync attribute-->
<attribute name="endsync" type="smil20:endsyncTimingAttrsType"/>

This definition also does not permit escaping the "media" etc. with a backslash. Which means that the definition of "endsync" in the ODF schema matches the definition in the SMIL XSD schema, but that the SMIL XSD schema does not match the SMIL prose. This is therefore an issue within SMIL itself. Even though the SMIL XSD schema is informative, it anyway seems to be reasonable to stay with the current definition. Another reason is that "escaped" IDs are not a concept of either XML or XSD. The impact of not allowing the four predefined endsync values as IDs is minor.

@Michael: +1

I concur. I checked the ODF 1.2 RNG Schema and we seem to be in the clear so long as choice lists always take the first winner. In that case, first, last, all, and media are reserved and will never be IDREFs.

No action.