Scheme explicitly assigns location tags to procedures for eqv? and eq? comparisons. Why isn’t this same treatment applied to procedures as it is implicitly for pairs, vectors, and strings?
i gotta think its about keeping the identity of functions clear without the overhead of structural comparisons, which is easy to mess up in more mutable types like pairs and vectrs. so its a design choice to focus on unique proc id’s for eq? comparisons.
Based on my observations, the distinct location identifier for procedures in Scheme is designed to ensure that the identities of functions are preserved without relying on structural content. Procedures, unlike pairs or vectors, represent executable code where only identity matters. This avoids the complexities and potential errors involved in structurally comparing code bodies. It simplifies comparisons by making them pointer-based, which is efficient and conceptually clean given that functions are first-class citizens that don’t naturally decompose into parts for meaningful structural equivalence.
I have found that the distinct procedure location identifier in Scheme provides an elegant way to handle function identity without getting entangled in the complexities of comparing actual code structures. This approach ensures that each compiled procedure has a unique reference, making equality comparisons fast and reliable by relying on pointer identity rather than structural equivalence. From my own experience in debugging and working with functional programming styles, I noticed that this design prevents ambiguous situations where two procedures might appear identical in structure but actually differ in their environment or execution context. It simplifies the reasoning about which function instance is being used.
i reckon its more about performance. each proc gets a uniq tag so you can cheaply check identity, avoiding messy deep structural comparisons that could bog down performance and create unexpected bugs.