Am Mo., 7. Sept. 2020 um 23:16 Uhr schrieb John Cowan <xxxxxx@ccil.org>:
>
>
>
> On Mon, Sep 7, 2020 at 2:31 AM Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:
>
>> <p>Unless otherwise noted, the procedures in this SRFI that return
>> ranges allocate at most O(1) new locations in the store. By abuse of
>> notion, such procedures are said to return <dfn>compact ranges</dfn>.
>> The procedures in this SRFI that do not guarantee to return compact
>> ranges allocate at most O(n) new locations where n is the length of
>> the resulting range. Again by abuse of notion, such procedures are
>> said to return <dfn>extended ranges</dfn>.</p>
>
>
> LGTM.  But there is no abuse of terminology here, as the terms compact and expanded are applied only to ranges.  (By contrast, the use of "a variable is bound to a value" in R[57]RS *is* an abuse of "bound", because strictly speaking a variable is bound to a location which is then assigned to a value.

The attributes compact and expanded may grammatically only be applied
to ranges, but that's logically incorrect (whether I wrote it is an
abuse of notion). Being compact and expanded are no properties that
can, logically, be ascribed to a range. One can see it when looking at
the procedures vector-range and vector->range. The latter can be
implemented in terms of the former: (define (vector->range v)
(vector-range (vector-copy v))). In particular, one cannot perceive
any difference between a range returned by the former and a range
returned by the latter. What differ are the procedures (being range
constructors) themselves. The former does not newly allocate O(n)
locations, the latter does.

Marc