Am Fr., 19. Aug. 2022 um 12:25 Uhr schrieb Lassi Kortela <xxxxxx@lassi.io>: > > > See the message at > > https://srfi-email.schemers.org/srfi-17/msg/2778561/ and the following > > discussion for why there are good arguments against a SRFI-17-style > > addition to Scheme. I definitely subscribe to these arguments. Per, > > the author of SRFI 17, does not. But we already had such a discussion > > and continuing it should happen somewhere else. > > There are two distinct questions here: > > 1) Should set! be able to set things other than variables? > > 2) Should references/places be first-class or implied? > > By my reading, your argument is that Scheme is cleaner (from a > functional programming standpoint?) if references/places are first class. A locative is a first-class object. A place would be, like a variable, not first-class. "set!" in Scheme is syntax and does not mutate its argument (which is some syntax) but the location which the syntactic argument references. "set-car!" (and friends) are procedures taking a first-class (evaluated) argument and mutating it. It's not about that all entities should be first-class. Variables aren't, for example. As I don't want to repeat all arguments, let me cite one I find compelling: https://srfi-email.schemers.org/srfi-17/msg/2778611/.
