>> Every SRFI document ever published has been under the MIT License, with
>> no exceptions; that license is obligated by the SRFI process.
>
> I'm suggesting that the process be rethought, because the MIT license
> is meaningless at best for a derivative work like srfi-170.html, and
> soon srfi-205.html.  I'm not sure if any harm can come from it's
> invalid assertion, given the below test that's before SRFI 170's BSD
> license, but I can't see any good coming from it.

Let's hear Arthur's opinion on this.

>> This SRFI is derived from the documentation for
>> <a href="https://scsh.net/">scsh</a>, whose copyright notice, from
>> the <code>COPYING</code> file, is reprinted here:
>
> That needs a tweak because we added my and John's name to the list of
> authors:
>
>> [...]
>> Copyright (c) 2001-2003 by Michael Sperber.
>> Copyright (c) 2019-2020 by John Cowan and Harold Ancell.

If we add authors, it's no longer reprinted verbatim so the comment
introducing the the license notice needs to be edited.

>> Presumably the SRFI 170 document contains enough text from the scsh
>> manual that the extra BSD notice is needed to cover those parts.
>
> In US copyright law, and I assume in general now that it's harmonized
> with the Berne Convention, it's a derivative work

That sounds about right.

>> Sample implementations don't have to be under the MIT license if code is
>> copied from an existing source. Do we have substantial code copied from
>> scsh? If so, the sample implementation's source tree needs a copy of the
>> BSD license....
>
> Yes; as it turns out, there's even more code from Chibi Scheme, and
> both use the same version of the BSD license, so we just added to the
> list of authors.

In this case, Alex has dual-licensed some of his BSD code under MIT, and
Olin was averse to legalese in his time (which can be witnessed in code
comments in sample implementations he wrote). It may be possible to get
the parts pertaining to SRFI 170 dual-licensed, simplifying the notices.

As far as I know, MIT and BSD have no material difference in rights and
restrictions, just different wording.

>> IMHO it would be useful to put a SPDX-License-Identifier at the top of
>> each source file in the sample implementation. That way it's clear which
>> license applies where....
>
> I'll look into this.

For an example, check any Scheme source file in
<https://gitlab.com/akkuscm/akku>.