R7RS-small defines alists, which are a longstanding data structure for key-value associations that has a reader syntax. I think that's why converting between hash tables and alists is important. (IIRC, that came up on another thread.) With those conversion routines you can convert a hashtable to an alist then write the alist, and vice-versa.

Kevin Wortman

On Fri, Sep 11, 2015 at 7:21 AM, Taylan Ulrich Bayırlı/Kammer <xxxxxx@gmail.com> wrote:

Per Bothner <xxxxxx@bothner.com> writes:

> On 09/11/2015 03:04 PM, John Cowan wrote:
>> Taylan Ulrich Bayırlı/Kammer scripsit:
>>
>>> Still, I guess I'll make it optional, along with weak/ephemeral
>>> hashtables, so there's still some utility of the SRFI for those who
>>> don't like all of it.
>>
>> I have no problem with making it optional. But I suggest that in that
>> case you adopt Racket's lexical syntax rather than a minor change from
>> it, so there will at least be some implementation.
>
> I agree.
>
> A bonus is if hashtables take part in quasi-quotation, so
> you could use the "same" syntax for computed values:
>
> `#hash((,n1 . ,v1) (,n2 . ,v2))
>
> In the above n1 v1 n2 v2 are computed expressions.
>
> Regardless, the result should (probably) be an immutable hash-table.

Indeed, I just pushed a change to my live version that specifies
quasiquote's behavior on hashtables.

https://github.com/TaylanUB/srfi-126/blob/master/srfi-126.md#quasiquote

(Note that I frequently force-push to that repo, until the next draft is
stabilized.)

Taylan

To unsubscribe from this list please goto http://www.simplelists.com/confirm.php?u=UfUPTfQ3psTelVWs9NbsMTTIEl1uPGRi