On Tue, Jul 4, 2017 at 2:22 AM, Marc Nieper-Wißkirchen <xxxxxx@nieper-wisskirchen.de> wrote:

1) A way to create sets using comparators that only provide an order predicate (because code using SRFI 113 in a portable fashion will have to provide both an order predicate and a hash function because the underlying implementation is opaque).

2) Explicit immutability of sets.

Correct, except that I would word (1) in terms of being able to take advantage of having an order predicate so as to process the set in order, which SRFI 113 does not portably allow.

If we have explicitly immutable ordered sets, there is no reason not to include explicitly immutable mappings with ordered keys

The (srfi 142) library already offers such a guarantee of ordering, or the ordering-dependent functions won't work.

yielding a new SRFI (combining SRFI 146 and SRFI 153).

Sets and mappings were once in a single pre-SRFI, and Arthur said it was too big to review, with which I agreed in retrospect. So I do not want to see them combined again.

Furthermore, there are legitimate use cases for immutable mappings based on (persistent, functional) hash tables.

To answer my own question from the other thread, there seems to be no persistent mapping implementation that has O(1) retrieval, less than O(n) storage.

So we would create just another SRFI for explicitly immutable mappings with hashed keys.

Simple immutable hash tables with O(n) mutation time (that is, by copying) are already optionally provided by SRFI 125. The sample implementation does provide them unless it is built directly on an implementation's own SRFI 69 (I think).

But what users will care about, I believe is: (1) Is cheap and safe mutation guaranteed? and (2) Is ordering guaranteed? With this SRFI we now have:

Guaranteed for sets: SRFI 153

Guaranteed for mappings: (srfi 146)

Not guaranteed for sets: SRFI 113

Not guaranteed for mappings: (srfi 146 hash)

So my question is: Why don't you make SRFI 153 spec-wise a thin layer over SRFI 113

Primarily because of the post-finalization errata of SRFI 113, which needed to be incorporated into SRFI 113.

John Cowan http://vrici.lojban.org/~cowan xxxxxx@ccil.org

I must confess that I have very little notion of what [s. 4 of the British

Trade Marks Act, 1938] is intended to convey, and particularly the sentence

of 253 words, as I make them, which constitutes sub-section 1. I doubt if

the entire statute book could be successfully searched for a sentence of

equal length which is of more fuliginous obscurity. --MacKinnon LJ, 1940