Dear Arthur,

Please withdraw SRFI 256 with the following notice:

> While this proposal achieves its semantic goals, no known Scheme implementation, including otherwise highly-optimizing ones, is yet able to optimize record construction in the style proposed here for subtype instances in all cases. I am continuing to work on the problem of finding a record type design that will satisfy Schemers’ desires for an efficient basis for a record system which can be extended to support any of the features someone might want.

And for those who are interested in what my current idea looks like: <https://gitlab.com/dpk/presrfis/-/blob/master/turtle-records.md>

Daphne