Am Fr., 11. Sept. 2020 um 18:58 Uhr schrieb John Cowan <xxxxxx@ccil.org>:

> On Fri, Sep 11, 2020 at 11:56 AM Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:
>
>> That something had to be done was probably inevitable, and indeed R7RS
>> fixes some flaws of R6RS. But such a fix could have been much more
>> conservative with respect to R6RS.
>
>
> If you read the WG1 and WG2 charters at scheme-reports.org, you'll see that that was *never* the intent. I actually tried to put stronger R6RS compatibility language into the WG2 charter, and the Steering Committee turned it down.  Will Clinger, after all, took his name off the R6RS report because he couldn't stand what it had turned into.

It may not have been the intent according to WG1 and WG2 charters, but
that doesn't mean that a more conservative approach couldn't have been
possible and that, in hindsight (!), the charters were optimal or
nearly optimal.

That said, I don't think that things are irreconcilable. There is
enough room for an R8RS, which closes the bridge between R6RS and R7RS
again. Such an R8RS could be seen as a further development of R6RS,
R7RS (small) would be (isomorphic) to some core of it and R7RS (large)
could be built on it.

>> Of course, hindsight is easier than foresight, but from
>> today's perspective, things weren't handled optimally after R6RS had
>> happened.
>
>
> I think we got just enough R6RS into R7RS-small, and there is already some R6RS in R7RS-large and I suspect there will be more.

From a technical point of view, probably true (although we still have
to see what will go into R7RS-large). I spoke about the political
side, though.

>> But the
>> people behind Chez seemed to have stopped working in favor of new
>> standards after the split after R6RS.
>
>
> True.  Then again, there is MIT Scheme, which doesn't even have modules (and is most certainly in active use).  Non omnia possumus omnes.
>
>> The same is true for the Racket
>> people.
>
>
> Racket has R6RS support, but it's not well integrated: I would go so far as to say its use is discouraged.

They once said that they would continue to support it (as they
continue to support R5RS). Judging from the "R6RS is 'perfect'" letter
by Matthias Felleisen, I suppose that, initially, there were plans to
keep closer to the development of the Scheme standard. But that's just
my guess.

>> The people that are currently active in the SRFI process, we,
>> are just a small circle, far away from the academic circles that had
>> defined and shaped Scheme.
>
>
> That's true.  But you underestimate the knock-on effects of the Steering Committee.

What do you mean exactly?

>> If I look at what they are doing, I can only take off my head to them.
>> While I may or not may like everything they put into Racket, they do
>> invent new things and develop the language further (on a certain scale
>> much more than we do here as we do mostly trivial things).
>
>
> That is the proper role of an implementation, and we do take things from implementations, including R6RS ones.  Standardizers should, strictly speaking, *invent* nothing.  I have often gone beyond that, but even so I have done my best to extend things as conservatively as possible.  R6RS, on the other hand, was full of inventions: some good, some not so good.

I agree. But even when "just" standardizing things, it is immensely
helpful if one is able to invent and develop and implement and think
about such things as these people are. To be honest, SRFIs enter R7RS
(large) through a voting system but I doubt that many voters do really
understand the technical subtleties or implications about the things
they are voting in favor or against. Nor do (or can) they care about
implementability. And I don't want to exclude myself. (That's actually
one reason why I have been thinking of abstaining any further votes
because exaggerated, democracy is good for society, but not for
science.)