schemaComplete attribute is currently defined as use=required (no default). I believe the intention behind this decision was to require the SML-IF producer make an explicit declaration regarding the completeness of the schema set. This is a good intention, but unfortunately this results in breaking all existing SML-IF files since they do not specify the schemaComplete attribute.

The WG discussed semantics of schemaComplete and decided on the following:

schemaComplete=true  : schema set is known to be complete
schemaComplete=false : schema set is not known to be complete

(see bug# 4774 for a discussion of the above.)

Note that, only the true value above is an explicit assertion. It asserts that schema set is known to be complete. On the other hand, the false value merely indicates the absence of an assertion.

Proposal:
Given the above meaning, it would be reasonable to have schemaComplete=false as the default value. 

This has 3 advantages,
1. It applies correctly to the existing models because we do not know the schemaComplete-ness of the existing models. Having this default minimizes the impact on the current models.
2. It retains the explicit declaration requirement for the true case where it makes most sense.
3. We do not have to change the meaning of true or false values.