Email list hosting service & mailing list manager

istring? Marc Nieper-Wißkirchen (15 Jul 2016 12:36 UTC)
Re: istring? Per Bothner (15 Jul 2016 14:25 UTC)
Re: istring? John Cowan (15 Jul 2016 15:53 UTC)
Re: istring? John Cowan (15 Jul 2016 14:59 UTC)
Re: istring? Per Bothner (15 Jul 2016 15:47 UTC)
Re: istring? John Cowan (15 Jul 2016 16:18 UTC)
Re: istring? Marc Nieper-Wißkirchen (16 Jul 2016 13:19 UTC)
Re: istring? John Cowan (16 Jul 2016 15:14 UTC)

Re: istring? John Cowan 16 Jul 2016 15:14 UTC

Marc Nieper-Wißkirchen scripsit:

> How many Schemes have already expressed their intent to implement
> R7RS-large?

None, at present: that would be an excessively open-ended commitment
from anyone at this stage.  However, Larceny does implement the whole
of the proposed Red Edition except SRFI 140, and I wouldn't be surprised
if that situation continued into later editions.

> For the users of Scheme (and I guess most of the voters are not (or
> won't be) implementors of an R7RS-large Scheme), it is nice to have
> a lot of features included. But someone has to implement all this.

That's why I decided early on to leverage the SRFI process, with its
emphasis on implementations that are as portable as possible.  Right now,
any R7RS-small system can do everything in the ballot by using the SRFI
sample implementations, although some things like eq-based hash tables
and ephemerons will be less than optimal.  This also means that users
can adopt these SRFIs themselves without waiting for implementers.

Things will become more tricky in the Green Edition, which will
(if current plans are followed) handle things that cannot readily
be implemented portably.  But hopefully by that time we will have
implementers willing to handle these things.

> Scheme seems to be a schizophrenic in this regard: Implementations
> are allowed to implement ephemerons using strong references because
> garbage collection is not observable. In the other hand, the
> standard demands that implementations are properly tail-recursive.
> However being non-properly tail-recursive is as non-observable as
> not implementing proper ephemerons.

An excellent insight.

--
John Cowan          http://www.ccil.org/~cowan        xxxxxx@ccil.org
Real FORTRAN programmers can program FORTRAN in any language.  --Ed Post