On Wed, 28 Oct 2009 21:12:34 -0400, Brian Harvey <bh@cs.berkeley.edu>
wrote:

> In the section "Requirements and Goals," the sentence
>
> 	A semantics compatible with interactive read/eval/print loops should
> 	be provided.
>
> should, I believe, be replaced with
>
> 	The semantics of the language must be compatible with interactive
> 	read/eval/print loops.

Since I believe that subset/superset nature of the two languages is of
primary importance (that is, having still a single language called
Scheme), I must express my opposing opinion here, since the above
requirement would necessarily limit the WG2 language far too much.

> I don't know whether this was carefully thought out by the SC or just an
> accident, but I'm disturbed by the fact that the discussion of
> compatibility
> with R5RS and other traditional Schemes under "Requirements and Goals"
> uses
> "should," whereas the discussion of compatibility with WG2 (\approx R6RS)
> under "Coordination with WG2" uses "must."  In my opinion this gets the
> importance of the two goals backwards, and also presupposes that this
> goal
> is compatible with the others, which some of us don't believe.  I would
> prefer to see "should" language in this section also.  ("Should if
> possible"
> would be even better!)

Because of my reasoning above, which we (Brian and I) have already
discussed to some degree, I disagree here, and agree with the existing
charters that the "must" requirement is a more important one than
maintaining absolute backwards compatibility with R5RS.

> Under "Membership," does "should endorse the goals of the working group"
> mean that agreeing with the "must" about WG2 compatibility is a
> requirement
> for membership?  If not, what /does/ it mean?  And if so, I think this is
> a recipe for leaving the people who ended up feeling disenfranchised by
> R6
> still feeling disenfranchised by R7.

Most of the disenchantedness I saw from the R6RS was a complaint against
increased complexity of the language on the library side, with some
handling of datatypes such as strings brought in. I don't believe the vast
majority of Scheme users had a problem with R6RS as its absolute
minimalistic core sans the many libraries and incompatibilities it
introduced. The language used to clarify the language to the extent
necessary given the increases in functionality (procedural macros) can
very easily be elided in WG1 such that the expansion is compatible with
WG2 while avoiding any undue complexity. However, this has already been
discussed, and I'm just reiterating this here as a more focused
counter-opinion, not to start this debate up again. We can do that
somewhere else.

	Aaron W. Hsu

--
Of all tyrannies, a tyranny sincerely exercised for the good of its
victims may be the most oppressive. -- C. S. Lewis

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports