Three suggestions for the text of SRFI-260
Jeronimo Pellegrini 28 Jan 2025 20:11 UTC
I have some suggestions, but none of them is really essential:
1. Perhaps it should be clearer that the "pretty-name" argument can be
ignored
by the implementation (I know this is what the reference
implementation does,
but it would be nice to state it in the SRFI text):
"The implementation may use the argument pretty-name for debugging
or
pretty-printing purposes, or it may just ignore it"
2. Although the "Implementation" section of the text mentions random
numbers
and UUIDs, the specification of the procedure could perhaps also be
clearer
as to what "for all practical purposes" means, Maybe "(that is, with
very high
probability)"?
3. Maybe mention that generate-symbol does *not* create uninterned
symbols. And
explicitly say that the symbol that is generated is can be referenced
by the
generated string that is its name (I know this is one of the examples
given,
but perhaps it would be nice to have it said explicitly?)
Just to avoid mistaken implementations like
(define generate-symbol gensym)
where gensym creates uninterned symbols.
J.