Do we support the near ubiquitous SRFI-23 error procedure?
Note that SRFI 23 error is incompatible with R6RS, because in R6RS the first argument is the object associated with the error (the "who"), whereas in SRFI 23 the first argument is the error string and all the rest are irritants.
We could have two procedures, error and error-on or the like.
I'd say that we stick with the R6RS error function. It's more useful, and the SRFI-23 ERROR is trivial to implement in the R6RS ERROR function.
The WG voted to accept SRFI-23 error.