Mixing Schema-aware and unaware XSLTs in XProc

Hello,

Suppose I have a mixture of schema-aware and unaware XSLTs that I use in an XProc pipeline. When I feed a PSVI-annotated XML input (i.e. the results of p:validate-with-xml-schema) to a schema-unaware XSLT, some interesting things happen. Data types change, causing some runtime errors. For example, xsd:ID types are recognized, causing the use of tokenize<https://github.com/usnistgov/cdf-test-method/blob/5078adb00df8bec2274a1373c4819de4fc6c7e98/resources/err_v2/err_v2_sua.sch#L26> to throw an error such as

Error code: err:XPTY0004 Reason: A sequence of more than one item is not allowed as the first argument of fn:tokenize() ("prec-1-2A", "prec-5-2")

My preference is to rewrite (if possible) the SUA XSLTs so they will work even if a PSVI if provided. The question is, how?

John Dziurłaj /d͡ʑurwaj/