> PS I think your model would work if SRFIs only documented well-tested
> and widely supported features.

That would greatly improve the SRFI process, and would save RnRS.

We submit experimental stuff because there isn't another place to get
code review and publicity. Not because it makes sense intrinsically.

>> SRFIs*do*  get bug fixes.  As is currently the case, they are not frozen.

Some bug fixes are fine. See my reply to John.

>> SRFIs are not libc, anyway.

SRFI is like if libc were factored into sub-libraries that you could
take or leave.

The early standard libc inherited many Unix features that hadn't stood
the test of time. Those keep being a burden now, decades later.