Hi, I’ve rewritten the sample implementations of bitvector-unfold and of -unfold-right. As Marc pointed out to me, the simple approach to these procedures allows still-to-be-mutated results to leak out if multiple returns occur. The new versions simply unfold a list and then convert it with list->bitvector.
Unfortunately, the SRFI doesn’t require this referential transparency. As with R7RS vector-map, results of earlier returns shouldn’t be mutated. bitvector-map and friends are unaffected in this implementation, but I think SRFI 133 might need some attention. Regards, Wolf Pull request: https://github.com/scheme-requests-for-implementation/srfi-178/pull/15 -- Wolfgang Corcoran-Mathe <[email protected]>
