I have decided to keep the names make-sum-comparator and make-product-comparator and finalize the SRFI with them. I have made the ‘warning’ about the name slightly more extensive for the very last draft, which waiting for Arthur to merge.

My full reasoning is:
1. There is precedent for the name make-sum-comparator in Schemepunk
2. Having accepted that precedent, the natural name for the complementary procedure is make-product-comparator
3. This reflects what 99% of what people will use it for in practice, even if theoretically unsound
3a. (Side point: I’m less and less convinced that the name even is theoretically unsound)
4. If someone were to create a ‘true’ product type comparator constructor in Scheme, it would make most sense to use separate comparators for the individual components of the type anyway, because you need to know the appropriate equality and ordering conventions and hash functions for the types used in each slot. Since you have make-wrapper-comparator anyway, it doesn’t make sense to duplicate that functionality within the product type comparator constructor; it does actually make practical sense to reuse make-wrapper-comparator for the mechanism which breaks a product type down into its components. Thus, a ‘true’ product type comparator constructor doesn’t make practical sense, so this will not conflict with any future attempt to define such a ‘true’ composition procedure.

Daphne