Poll for next WG2 meeting

[Wolfgang Corcoran-Mathe]

Hello all,

Thanks to everyone who attended today's meeting. I'll try to post
the minutes shortly. I'm getting out ahead on the poll for the next
one:

https://terminplaner6.dfn.de/p/ace0d1db22d15c9d0f9094bb5e04c441-1708506

Please fill in your available times.

Thanks,

Wolfgang

--
Wolfgang Corcoran-Mathe <w...@sigwinch.xyz>

[Alaric Snell-Pym]

On 16/04/2026 16:26, Wolfgang Corcoran-Mathe wrote:
> Hello all,
>
> Thanks to everyone who attended today's meeting. I'll try to post
> the minutes shortly. I'm getting out ahead on the poll for the next
> one:
>
> https://terminplaner6.dfn.de/p/ace0d1db22d15c9d0f9094bb5e04c441-1708506

I've already filled it in, but then since found I'm the closest next of
kin to somebody who died alone at home, so at some point over the next
weeks all my plans and schedules will be upended by needing to visit
another city to look for a will / clear a house / organise a funeral /
other paperwork... Hopefully I'll be able to bring my laptop and get
Internet access, but just to warn you, I might be a bit awkward! Sorry!

>
> Wolfgang
>

--
Alaric Snell-Pym (M0KTN neé M7KIT)
http://www.snell-pym.org.uk/alaric/

[Wolfgang Corcoran-Mathe]

Alaric,

On 2026-04-18 12:17 +0100, Alaric Snell-Pym wrote:
> I've already filled it in, but then since found I'm the closest next
> of kin to somebody who died alone at home, so at some point over the
> next weeks all my plans and schedules will be upended by needing to
> visit another city to look for a will / clear a house / organise a
> funeral / other paperwork... Hopefully I'll be able to bring my
> laptop and get Internet access, but just to warn you, I might be a
> bit awkward! Sorry!

Thanks for the update. I'm sorry this has happened and I understand
you'll be busy tracking down a thousand details in the coming weeks.
Would you like the rest of us to come up with an agenda, or select
an acting^2 chair for the next meeting?

--
Wolfgang Corcoran-Mathe <w...@sigwinch.xyz>

[Peter McGoron]

Re. balloting, I have put down some questions that are related to the
drafting of the Valued Fascicle:

https://codeberg.org/scheme/r7rs/wiki/The-Aquamarine-Ballot

Some important things:

1. Decide on what we should ask for uninterned symbols. Explicit library
support? Implicitly allowed like in R7RS-Small? Forbidden? (Maybe we can
ask all three and use ranked-choice voting.)

2. Further Unicode scope questions, as necessary.

3. How much of IEEE arithmetic to require, and if we want to require
flonums to be IEEE binary64. (This would also touch upon requirements
for using the most precise floating point format available, R6RS §3.4
paragraph 5).

-- Peter McGoron

[John Cowan]

On Mon, Apr 20, 2026 at 7:47 PM 'Peter McGoron' via scheme-reports-wg2
<scheme-re...@googlegroups.com> wrote:

> https://codeberg.org/scheme/r7rs/wiki/The-Aquamarine-Ballot

Note to all: the version at this link is much longer than the version
in this email.

I have added "or both" to the question about #u1 syntax vs. #* syntax
for bitvectors, and by adding "or neither" we can eliminate question 1
(but I didn't do that).

> 3. [...] (This would also touch upon requirements

> for using the most precise floating point format available, R6RS §3.4
> paragraph 5).

There are some existing issues concerned with this: #67 (mixed
precision, which I just reopened to add a comment) and #275, which is
about requiring flonums to be binary64 and adding optional
C#-compatible syntax for (optional) decimal floats.

I'll add my views on this ballot in brief:. Bitvector syntax: both.
Mantissa width syntax: yes. Multidimensional array syntax: yes per
SRFI 268, producing a result that is opaque in the Foundations.
Uninterned symbols: no (I see no use cases for them any more).
Unicode scalar values only: yes. Locale support: maybe in the
Batteries. Strict IEEE 754 compliance: yes.

[Wolfgang Corcoran-Mathe]

We have a bunch of poll responses and it looks like next Thursday
(2026-04-30) at 13:00 UTC works for everyone (except perhaps Alaric,
whose schedule has become complicated).

Unless there are objections I'll schedule that as the next meeting
time.

Alaric, if you have a chance, please let us know if you need someone
to handle the agenda & chairing duties.

--
Wolfgang Corcoran-Mathe <w...@sigwinch.xyz>

[Peter McGoron]

On 4/20/26 20:46, John Cowan wrote:
> Mantissa width syntax: yes.

I have some doubts about balloting it now. We would have to modify the
syntax to make the grammar LL(k). If we add the R6RS mantissa width in
combination with escaped identifiers in the way the R7RS describes them,
then you need arbitrary lookahead to disambiguate a mantissa width from
the start of an escaped identifier. Example:

(let ((|111111111| 5))
(cons 5|111111111|))

The only correct parse of this is the one that evaluates to (5 . 5),
which would require the parser to back out of parsing 5 with a very long
mantissa. Another example:

(let ((|abcd| 5))
(cons 5|111111111|abcd|))

this would parse as (5.0|111111111 . 5), but would also require some
tricky parsing to get right.

______________________

I've added many more questions to the proposed ballot:

https://codeberg.org/scheme/r7rs/wiki/The-Aquamarine-Ballot

-- Peter McGoron

[Alaric Snell-Pym]

On 18/04/2026 16:45, Wolfgang Corcoran-Mathe wrote:
> Alaric,

>
> Thanks for the update. I'm sorry this has happened and I understand
> you'll be busy tracking down a thousand details in the coming weeks.
> Would you like the rest of us to come up with an agenda, or select
> an acting^2 chair for the next meeting?

Yeah, I think I'm going to need to sit this one out, sorry. Please
proceed without me!

(It's taken nearly two weeks for the police to actually respond to my
requests for information - I was notified by my local police coming to
my home at just past midnight on the Friday before last, and they
collected my details as next-of-kin to pass on to the police force in
the different county that found her, so they could get in touch with me.
After two weeks of me ringing and emailling them, and even driving
across the UK to visit the station in person, all I'd got was promises
they'd get back to me as soon as they could. Well, this morning, they
finally did! And told me a little more information, but still couldn't
tell me when I could come and start to deal with things at the house!
But they need to ring me back to tell me when, and it might be tomorrow!)

Equal parts frustrating and stressful, I'm afraid :-( I'm not falling
apart, personally - I'm just struggling to think about anything else,
and struggling to commit to any schedules as I want to go and start
dealing with things as soon as I possibly can. I should be back and
eager to help in another week or two!

Thanks for your patience,

[Peter McGoron]

That's awful. I hope that all gets sorted out.

I can come up with the agenda and be acting² chair, if you and everyone
else is OK with that.

-- Peter McGoron

[Alaric Snell-Pym]

On 29/04/2026 14:47, 'Peter McGoron' via scheme-reports-wg2 wrote:
> That's awful. I hope that all gets sorted out.

Me too!!!

> I can come up with the agenda and be acting² chair, if you and everyone
> else is OK with that.

Thanks Peter :-)

>
> -- Peter McGoron

[Peter McGoron]

Hello all. Here is the agenda:

https://codeberg.org/scheme/r7rs/wiki/Agenda+2026-04-30.-

See you there!

-- Peter McGoron

[John Cowan]

Feedback, just in case:

IEEE math: I would prefer conformant behavior (your option 1), but at
least require flonums to be binary64 (your option 2), since that is
provided by every Scheme and every CPU that has floating-point support
at all. Allowing other IEEE formats (your option 3) as flonums seems
like an unnecessary freedom, although allowing them as alternatives is
fine. There are no non-IEEE systems with infinities or negative
zeros other than those constructed on IEEE principles but not (yet)
standardized by IEEE, so your option 4 makes little sense. I don't
know what mixed-exactness complex numbers have to do with the other
questions.

Funny symbols: I see no point in SRFI 258 truly uninterned symbols;
the SRFI 260 elements with unpredictable (and lazy) names suffice.

utf8->string: strict behavior only.

normalization: R6RS is sufficient.

mixed-exactness: remove the language about always using the highest
available precision when returning an inexact value from exact values.

optional start-end on list-copy (not useful enough) and substring
(currently the same as string-copy but with required arguments.
Status quo.

gcd on rationals: no strong opinion, little existing support.

domain of number predicates: extend all of them to accept any object.

hex lloats: support @phm's version

equal?: a question of wording, no strong opinion.

non-Unicode characters: eliminate.

member/assoc: no opinion yet.

conjugate: in favor. C, C++, Fortran, Algol 68, Ruby, Julia. Haskell,
Golang all support this; only the minimal support in Python and
sub-minimal support in Rust lack it.

> --
> You received this message because you are subscribed to the Google Groups "scheme-reports-wg2" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to scheme-reports-...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/scheme-reports-wg2/4fb08407-01ad-497a-b64a-6626fe3cde53%40mcgoron.com.

[Peter McGoron]

On 4/30/26 01:03, John Cowan wrote:
> There are no non-IEEE systems with infinities or negative
> zeros other than those constructed on IEEE principles but not (yet)
> standardized by IEEE, so your option 4 makes little sense.

This is "make the optional parts of the R6RS required, but no more." I
will clarify.

> I don't know what mixed-exactness complex numbers have to do with the
other questions.

That is a separate issue so I will move it.

-- Peter McGoron