[r6rs-discuss] [Formal] Rationalize Iteration

[Anton van Straaten]

MichaelL at frogware.com wrote:
>  * Why is folding the only iteration offered for hash tables? 
> 
> Chez Scheme and PLT Scheme both offer hash table map and "each" 
> procedures. Though they could both be written in terms of hash-table-fold,
> R6RS follows a funny line, dramatically expanding the number of list 
> utilities while at the same time keeping a minimal set of operations for 
> other types--more or less. 
...
> Recommendations:
...
>  * If the philosophy is to provide a reasonable set, some thought should 
> be given to what that set should be, and that set should be implemented 
> for all types where it makes sense. A reasonable set might be fold with 
> premature termination, fold without premature termination (for 
> convenience), map, and for-each. 

Just an informal & narrow reply specifically about hash table map:

The map operation has an obvious, useful specification for lists and 
vectors: the former maps lists to lists, and the latter maps vectors to 
vectors.  One might reasonably extend this to hash tables and specify 
that a hash table map function should return a hash table.  This is what 
SML/NJ does, for example.

However, I'm not aware of any Scheme which does this.  According to SRFI 
69, "In no implementation that I know of, hash-table-map does a real 
functorial map that lifts an ordinary function to the domain of hash 
tables."  SRFI 69 omits hash table map for this reason.  Some Scheme 
implementations, such as Gambit and Chicken, implement hash tables but 
don't provide a map function for them.  Similarly, hash table libraries 
for some other languages, such as OCaml and Haskell, also omit map.

A number of other Scheme implementations, including Chez, PLT, Larceny, 
and Gauche, provide a hash table map function which returns a list, 
rather than a hash table.  This seems to indicate that a hash table map 
which returns a hash table is not considered particularly necessary in 
practice, which argues against specifying such a function in R6RS.

Ignoring a multitude of options that I haven't mentioned above, this 
leaves the following question: assume the editors decide, belatedly, 
that the R6RS philosophy should be "to provide a reasonable set [...] 
and that set should be implemented for all types where it makes sense", 
does it make sense for R6RS to specify a hash table map function which 
returns a list?

BTW, I don't consider any of this to affect the overall point of the 
original formal comment.

Anton