[intro.refs], [support] Remove all references to ISO/IEC 10967-1:2012

[eisenwave]

We have a bibliography reference to ISO/IEC 10967-1:2012 and mention it in a number of footnotes. That standard makes factually incorrect statements and is irrelevant to C and C++ following the C23 rebase. We should purge any mention of it.

Historical background

Historically, C17 had an Annex H which documented compliance with ISO/IEC 10967–1. I suppose this is why we refer to it as well. C23 no longer mentions ISO/IEC 10967–1 in any of its wording, but does contain an unused reference in its bibliography. This is most likely an editorial mistake.

C23 instead has an Annex F, G, and H that document compliance with ISO/IEC 60559. We also refer to ISO/IEC 60559, and we are based on C23 for large portions of the standard library.

The problem with referencing ISO/IEC 10967-1

ISO/IEC 10967-1:2012 is now essentially irrelevant to us. In fact, the references we have are outright wrong and misleading:

draft/source/support.tex

Lines 1659 to 1663 in ef2abfa

	The rounding style for the type.
	\begin{footnote}
	Equivalent to \tcode{FLT_ROUNDS}.
	Required by ISO/IEC 10967-1:2012.
	\end{footnote}

C23 (https://cigix.me/c23#5.2.5.3.3.p23) says that an FLT_ROUNDS value of 1 indicates "to nearest, ties to even" and a value of 4 indicates "to nearest, ties away from zero"

On the contrary, ISO/IEC 10967-1 claims:

This is wrong. C no longer has any ambition of complying with ISO/IEC 10967-1, and it outright contradicts it, as can be seen.

Note

The "†" symbol means:

The operations or parameters marked “†” are not part of the language standard and must be provided by an implementation that wishes to conform to LIA-1.

This is also wrong in the case of C23. The language standard does provide the value 4.

Thankfully, no normative wording is affected by this, only a bunch of footnotes in [support].

[jwakely]

SG6 should be asked to give an ack.

[jensmaurer]

If LWG (or the chair thereof) is happy with the direction, I'm ok with cleaning out the footnotes. However, excising all references to another standard (as misguided as it may be) feels a bit much to do by the editors autonomously.

I'd suggest you send a message to both the -lib and SG6 reflectors to ask for opinions.

[eisenwave]

I've sent an email to [sci] and [lib] a month ago (http://lists.isocpp.org/sci/2025/12/0878.php). While there weren't a whole lot of responses, there was no objection at least.

Is that sufficient, or do we need more feedback? I will inevitably be in SG6 if SG6 meets in Croydon, so we could get more feedback there.

CC @mattkretz

[jensmaurer]

Great, thanks. @mattkretz as chair of SG6, do you want to formally discuss this in SG6, or should we just go ahead editorially?

[mattkretz]

Sorry for my long silence. My gut feeling is that nobody knows about LIA-1 anyway. Those who would know are possibly in other groups (if any). If you want some form of visibility in the committee about this change, I think the minimum would be to go via LWG issue (and that way it passes plenary). I don't have a strong opinion either way, though.

[eisenwave]

Are we good to merge the change then? This sounds to me like a "no" to the question of whether SG6 wants to formally discuss this.

I did contact LWG in https://lists.isocpp.org/lib/2025/12/34358.php in addition to the SG6 reflector, as mentioned above. While opening an LWG issue is a possibility, no one insisted on one so far, and the change here is obviously editorial (just removing some footnotes). That would make me feel a bit silly when filing an LWG issue at this point.

[jensmaurer]

@jwakely , are you good with the level of socializing we've done?

[jwakely]

Late December isn't the time when people are most active on the reflectors, but people did get a chance to object and no objections were made. I think the change is fine (and I've added my approval to the PR now).