There are different XSLTs that need to be applied depending on whether the schema is "min" or "normal" (or both or neither). It is awkward to handle this increasing complexity through separate directories for the different classes.
Option 1: perform all the processing, from configuration to post-processing, in one pass.
Option 2: introduce an annotation that tells the (unique) post-processing XSLT how to post-process.
There are different XSLTs that need to be applied depending on whether the schema is "min" or "normal" (or both or neither). It is awkward to handle this increasing complexity through separate directories for the different classes.
Option 1: perform all the processing, from configuration to post-processing, in one pass.
Option 2: introduce an annotation that tells the (unique) post-processing XSLT how to post-process.