On 6/21/26 13:20, Wolfgang Corcoran-Mathe wrote:
> I think these are useful, but slightly outside of SRFI 271's scope.
> There is room for further random-port SRFIs.

The thing is, trivial becomes less useful if one defines a port to
generate values from a certain algorithm. If someone wrote a “specific
random ports” SRFI/library containing (say) ChaCha20, then I know that
ChaCha20 is seeded from a 256 bit key, so I can just feed it that fixed
amount of data. If I want to seed another algorithm, then I can just
seed ChaCha20 with my random data, and then seed the other algorithm
with ChaCha20.

The SRFI lacks a base-case that one can use to seed random ports from
known binary data. ChaCha20 is the base case in my hypothetical library,
while `trivial` is a base case that is also much simpler. The only real
use of `trivial` is seeding other random ports, which is why it should
be in this SRFI, which is the one that defines random ports and how
random ports can seed other ports.

-- Peter McGoron