Similar to #276 we know a priority which classes we can and will list in embedding and referencing. Now, we are using class names. We already apply some tricks to reduce the entropy with the identifier but we may reduce this much further just by using integers up to a certain number of bits. Even if it would be a 32bit value, that would at most compare 4 characters.
We should approach this from a list of classes (already available) being in a stable sorted order. It must be extremely stable if we want to have it working over different XML Schema versions, but if that would be ever a problem, it could also be stored as mapping in the database.
Similar to #276 we know a priority which classes we can and will list in embedding and referencing. Now, we are using class names. We already apply some tricks to reduce the entropy with the identifier but we may reduce this much further just by using integers up to a certain number of bits. Even if it would be a 32bit value, that would at most compare 4 characters.
We should approach this from a list of classes (already available) being in a stable sorted order. It must be extremely stable if we want to have it working over different XML Schema versions, but if that would be ever a problem, it could also be stored as mapping in the database.