On Fri, Oct 13, 2017 at 11:58 AM, Donald Allen <xxxxxx@gmail.com> wrote:
I must say that it's disconcerting to be finding all these problems in
code that was finalized in 2004.
Note that it is the SRFI prose, not the code, that is finalized. The code is just sample code, and may have bugs like any other code. Indeed, the requirement to provide tests is relatively recent and has not been made retroactive.
Nor is there any guarantee that the sample code will Just Work on all Schemes. Indeed, I just found a bug in my SRFI 158 test suite depending on the fact that (/ 8) can evaluate to either 0.125 or 1/8, and that these are not the same in the sense of `equal?`.