vmanis at telus.net
Tue Mar 24 20:37:02 EDT 2009
On 2009-03-24, at 17:16, Sam TH wrote:
> Unless "we" and "where I come from" is intended to imply something
> very narrow (ie, not including some large set of people associated
> with Scheme), then these are very strong claims about what a wide
> variety of opinionated people thought and intended and agreed upon in
> the past. Do you have any evidence for these claims? For example,
> can you provide any instances of other people propounding your
> "definition" of Scheme?
This whole discussion is making me extremely uncomfortable. In fact,
my head exploded when I read John Cowan's claim that `Scheme' is
devoid of operational or denotational meaning. I felt like a character
in a Philip K Dick story who discovers that not only is the universe
unreal, but so too is the character him/herself. I was also reminded of
the Stephen King story where the characters leave this plane of
existence; they discover that they are in an unreal universe by eating
airport cafeteria food and discovering it to be tasteless.
I know perfectly well what `Scheme' is. It's a somewhat fuzzy set of
languages that have reasonably common syntax,, semantics, and
pragmatics. It's quite practical to write portable Scheme code, which
must therefore mean that the language the author of the code is writing
in and the language the implementor is implementing must have a
very high degree of overlap. Indeed, Aristotle might object to `Scheme',
as there is no predicate that is true of objects that are members of
this set, and false of other objects. But as computer scientists, we
deal with fuzzy sets like this all the time (`programming language'
also has no such predicate).
I feel we have taken a real question about a gray area in the semantics
of string->number and turned it into an ontological debate. Personally,
I don't care which way a decision about what to do with invalid
to string->number goes. Saying that this case `is an error', or that an
implementation `should detect it', or that it's `unspecified' are all
with me. I can live with incompletely-specified languages (remember,
the internal representation of C integers are not specified, somebody
build a replica of an IBM 7094, and we might have sign-magnitude
integers again!). But this seems to me to be one of those places where
we make a decision, which might be to leave the wording as is, or to
change it in some way, and move on.
More information about the r6rs-discuss