C11/POSIX-2024 Questions

C11/POSIX-2024 Questions

[Joel Sherrill]

[-- Attachment #1: Type: text/plain, Size: 1345 bytes --]

Hi

I am finishing up reviewing the additions from POSIX Issue 8 which was
released recently. I noticed a few things and have questions on the right
approach.

+ I only have the final draft for POSIX-2024. It says "_POSIX_VERSION to
the value
20yymmL." It isn't on the web yet for the HTML version. Anyone know the
value this has in the final version?

+ sys/features.h appears to need this new POSIX version added at least to
the comment block. I don't think it necessarily changes any logic below
that.

+ memmem is in string.h under the __GNU_VISIBLE guard. Does it now need to
be under a __POSIX_VISIBLE > 2024mmdd and __GNU_VISIBLE?

+ aligned_alloc in stdlib.h is only guarded by C > 2011. Does that need to
be an || POSIX >= 2024 also?

+ C11 and POSIX-2024 add these complex methods.

double complex CMPLX(double x, double y);
float complex CMPLXF(float x, float y);
long double complex CMPLXL(long double x, long double y);

FreeBSD has a macro implementation (
https://github.com/lattera/freebsd/blob/master/include/complex.h#L49). If I
add that to complex.h, does it also need a __ISO_C_VISIBLE >= 2011 ||
__POSIX_VISIBLE > 2024mmdd guard?

+ Should I continue to look for guards on things that are newly added in
POSIX 2024 and already in newlib?

That's it so far. Just picky stuff that I plan to address given some advice.

--joel

Re: C11/POSIX-2024 Questions

[John Scott]

[-- Attachment #1.1: Type: text/plain, Size: 1506 bytes --]

> I only have the final draft for POSIX-2024. It says "_POSIX_VERSION to the value 20yymmL." It isn't on the web yet for the HTML version. Anyone know the value this has in the final version?
I assume you're referring to Draft 4.1, which indeed just has the placeholder. I don't have the final version, but the latest "interim build draft" had 202405L. I expect that this has not changed since this corresponds to the month that it was sent out for publication.

> memmem is in string.h under the __GNU_VISIBLE guard. Does it now need to be under a __POSIX_VISIBLE > 2024mmdd and __GNU_VISIBLE?
ISO C has reserved symbols starting with "str" and "mem" in <string.h> to the implementation at least since C99, and POSIX affirms this, so technically it does not, and has never needed, a guard at all. It can be exposed even in strict ISO C conformance mode unconditionally. This extends to strlcpy, strlcat, stpcpy, and friends too.

> aligned_alloc in stdlib.h is only guarded by C > 2011. Does that need to be an || POSIX >= 2024 also?
POSIX 2024 *requires* C11/C17: if someone has _POSIX_C_SOURCE set to 202405L but __STDC_VERSION__ < 201112L, that's not a conforming build environment and it's basically an error on the part of the user. I therefore don't think this deserves special accommodation: a conditional on the ISO C version should be adequate.

Please note: I'm basically an amateur who lurks here and in the Austin Group and you should have healthy skepticism over my opinions here.

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 6270 bytes --]

Re: C11/POSIX-2024 Questions

[brian.inglis]

On 2024-06-28 08:37, Joel Sherrill wrote:
> Hi
> 
> I am finishing up reviewing the additions from POSIX Issue 8 which was released 
> recently. I noticed a few things and have questions on the right approach.
> 
> + I only have the final draft for POSIX-2024. It says "_POSIX_VERSION to the value
> 20yymmL." It isn't on the web yet for the HTML version. Anyone know the value 
> this has in the final version?

The Austin Group minutes say in:

	https://www.mail-archive.com/austin-group-l@opengroup.org/msg12711.html

"The IEEE and The Open Group pdf edition was published on Friday June 14th."

so I would expect POSIX-1.2004/IEEE Std. 1003-1(2024) to have 202406L date.

Another email states it could take another month for the HTML ~July 21.

> + sys/features.h appears to need this new POSIX version added at least to the 
> comment block. I don't think it necessarily changes any logic below that.

The latest draft normative references include ISO/IEC 10646:2020 (6th edition) 
so Unicode should be at least 13 dated 2020-03-06 and __STDC_ISO_10646__ at 
least 202003L.
This was amended by ISO/IEC 10646:2020/Amd1:2023 2023-07-17 so Unicode could be 
up to 15.0 dated 2022-09-09 with __STDC_ISO_10646__ 202209L.
There were no amendments for Unicode 14 2021-09-10 or 15.1 2023-08-28.
Future considerations expect Unicode 16 in Amendment 2:2025.

ISO/IEC JTC 1/SC2 WG2 is no longer at http://std.dkuug.dk/jtc1/sc2/wg2 
(NXDOMAIN) so old working documents are no longer available but has moved to 
https://www.unicode.org/wg2/ where more recent working documents are available, 
and hopes are that ISO/IEC standards can be stabilized with normative references 
to Unicode specs like UTS Unicode Technical Standards and UAX Unicode Standard 
Annexes.

> + memmem is in string.h under the __GNU_VISIBLE guard. Does it now need to be 
> under a __POSIX_VISIBLE > 2024mmdd and __GNU_VISIBLE?
> 
> + aligned_alloc in stdlib.h is only guarded by C > 2011. Does that need to be an 
> || POSIX >= 2024 also?
> 
> + C11 and POSIX-2024 add these complex methods.
> 
> double complex CMPLX(double x, double y);
> float complex CMPLXF(float x, float y);
> long double complex CMPLXL(long double x, long double y);
> 
> FreeBSD has a macro implementation 
> (https://github.com/lattera/freebsd/blob/master/include/complex.h#L49 
> <https://github.com/lattera/freebsd/blob/master/include/complex.h#L49>). If I 
> add that to complex.h, does it also need a __ISO_C_VISIBLE >= 2011 || 
> __POSIX_VISIBLE > 2024mmdd guard?

The same minutes linked above says:

	"The ISO/IEC [POSIX] ballot closes on June 28."

Also note that ISO/IEC 9899:2024 C is targeted for July 12, and JTC 1/SC 22/WG 
14 hope to get that passed by some picky ISO editors in time, or it will be 
cancelled for this year.

So look for ISO C >= 2024 and add changes, including docs and compiler options, 
and note 202X is becoming ambiguous so groups are using 202Y.

> + Should I continue to look for guards on things that are newly added in POSIX 
> 2024 and already in newlib?
> 
> That's it so far. Just picky stuff that I plan to address given some advice.

I am also waiting to see if this will be issued as SUSv5?

-- 
Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada

La perfection est atteinte                   Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter  not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer     but when there is no more to cut
                                 -- Antoine de Saint-Exupéry

Re: C11/POSIX-2024 Questions

[Joel Sherrill]

[-- Attachment #1: Type: text/plain, Size: 4153 bytes --]

Thanks everyone. I guess the broader question with multiple
standards being released close to each other, is do we need
a master plan? Or just fix things as they are spotted.

It seems there are a number of things to address this time.

--joel

On Fri, Jun 28, 2024 at 4:33 PM <brian.inglis@systematicsw.ab.ca> wrote:

> On 2024-06-28 08:37, Joel Sherrill wrote:
> > Hi
> >
> > I am finishing up reviewing the additions from POSIX Issue 8 which was
> released
> > recently. I noticed a few things and have questions on the right
> approach.
> >
> > + I only have the final draft for POSIX-2024. It says "_POSIX_VERSION to
> the value
> > 20yymmL." It isn't on the web yet for the HTML version. Anyone know the
> value
> > this has in the final version?
>
> The Austin Group minutes say in:
>
>
> https://www.mail-archive.com/austin-group-l@opengroup.org/msg12711.html
>
> "The IEEE and The Open Group pdf edition was published on Friday June
> 14th."
>
> so I would expect POSIX-1.2004/IEEE Std. 1003-1(2024) to have 202406L date.
>
> Another email states it could take another month for the HTML ~July 21.
>
> > + sys/features.h appears to need this new POSIX version added at least
> to the
> > comment block. I don't think it necessarily changes any logic below that.
>
> The latest draft normative references include ISO/IEC 10646:2020 (6th
> edition)
> so Unicode should be at least 13 dated 2020-03-06 and __STDC_ISO_10646__
> at
> least 202003L.
> This was amended by ISO/IEC 10646:2020/Amd1:2023 2023-07-17 so Unicode
> could be
> up to 15.0 dated 2022-09-09 with __STDC_ISO_10646__ 202209L.
> There were no amendments for Unicode 14 2021-09-10 or 15.1 2023-08-28.
> Future considerations expect Unicode 16 in Amendment 2:2025.
>
> ISO/IEC JTC 1/SC2 WG2 is no longer at http://std.dkuug.dk/jtc1/sc2/wg2
> (NXDOMAIN) so old working documents are no longer available but has moved
> to
> https://www.unicode.org/wg2/ where more recent working documents are
> available,
> and hopes are that ISO/IEC standards can be stabilized with normative
> references
> to Unicode specs like UTS Unicode Technical Standards and UAX Unicode
> Standard
> Annexes.
>
> > + memmem is in string.h under the __GNU_VISIBLE guard. Does it now need
> to be
> > under a __POSIX_VISIBLE > 2024mmdd and __GNU_VISIBLE?
> >
> > + aligned_alloc in stdlib.h is only guarded by C > 2011. Does that need
> to be an
> > || POSIX >= 2024 also?
> >
> > + C11 and POSIX-2024 add these complex methods.
> >
> > double complex CMPLX(double x, double y);
> > float complex CMPLXF(float x, float y);
> > long double complex CMPLXL(long double x, long double y);
> >
> > FreeBSD has a macro implementation
> > (https://github.com/lattera/freebsd/blob/master/include/complex.h#L49
> > <https://github.com/lattera/freebsd/blob/master/include/complex.h#L49>).
> If I
> > add that to complex.h, does it also need a __ISO_C_VISIBLE >= 2011 ||
> > __POSIX_VISIBLE > 2024mmdd guard?
>
> The same minutes linked above says:
>
>         "The ISO/IEC [POSIX] ballot closes on June 28."
>
> Also note that ISO/IEC 9899:2024 C is targeted for July 12, and JTC 1/SC
> 22/WG
> 14 hope to get that passed by some picky ISO editors in time, or it will
> be
> cancelled for this year.
>
> So look for ISO C >= 2024 and add changes, including docs and compiler
> options,
> and note 202X is becoming ambiguous so groups are using 202Y.
>
> > + Should I continue to look for guards on things that are newly added in
> POSIX
> > 2024 and already in newlib?
> >
> > That's it so far. Just picky stuff that I plan to address given some
> advice.
>
> I am also waiting to see if this will be issued as SUSv5?
>
> --
> Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada
>
> La perfection est atteinte                   Perfection is achieved
> non pas lorsqu'il n'y a plus rien à ajouter  not when there is no more to
> add
> mais lorsqu'il n'y a plus rien à retirer     but when there is no more to
> cut
>                                  -- Antoine de Saint-Exupéry
>

Re: C11/POSIX-2024 Questions

[brian.inglis]

POSIX is also ISO/IEC JTC 1/SC 22/WG 15 ISO/IEC 9945:2024 in DIS ballot.

Probably easier to look for changes from drafts and propose changes and 
comments, then wait for both ISO/IEC ballots to pass or ISO/IEC updates to be 
cancelled, before applying updates.

On 2024-06-28 16:01, Joel Sherrill wrote:
> Thanks everyone. I guess the broader question with multiple
> standards being released close to each other, is do we need
> a master plan? Or just fix things as they are spotted.
> 
> It seems there are a number of things to address this time.
> 
> --joel
> 
> On Fri, Jun 28, 2024 at 4:33 PM <brian.inglis@systematicsw.ab.ca 
> <mailto:brian.inglis@systematicsw.ab.ca>> wrote:
> 
>     On 2024-06-28 08:37, Joel Sherrill wrote:
>      > Hi
>      >
>      > I am finishing up reviewing the additions from POSIX Issue 8 which was
>     released
>      > recently. I noticed a few things and have questions on the right approach.
>      >
>      > + I only have the final draft for POSIX-2024. It says "_POSIX_VERSION to
>     the value
>      > 20yymmL." It isn't on the web yet for the HTML version. Anyone know the
>     value
>      > this has in the final version?
> 
>     The Austin Group minutes say in:
> 
>     https://www.mail-archive.com/austin-group-l@opengroup.org/msg12711.html
>     <https://www.mail-archive.com/austin-group-l@opengroup.org/msg12711.html>
> 
>     "The IEEE and The Open Group pdf edition was published on Friday June 14th."
> 
>     so I would expect POSIX-1.2004/IEEE Std. 1003-1(2024) to have 202406L date.
> 
>     Another email states it could take another month for the HTML ~July 21.
> 
>      > + sys/features.h appears to need this new POSIX version added at least to
>     the
>      > comment block. I don't think it necessarily changes any logic below that.
> 
>     The latest draft normative references include ISO/IEC 10646:2020 (6th edition)
>     so Unicode should be at least 13 dated 2020-03-06 and __STDC_ISO_10646__ at
>     least 202003L.
>     This was amended by ISO/IEC 10646:2020/Amd1:2023 2023-07-17 so Unicode could be
>     up to 15.0 dated 2022-09-09 with __STDC_ISO_10646__ 202209L.
>     There were no amendments for Unicode 14 2021-09-10 or 15.1 2023-08-28.
>     Future considerations expect Unicode 16 in Amendment 2:2025.
> 
>     ISO/IEC JTC 1/SC2 WG2 is no longer at http://std.dkuug.dk/jtc1/sc2/wg2
>     <http://std.dkuug.dk/jtc1/sc2/wg2>
>     (NXDOMAIN) so old working documents are no longer available but has moved to
>     https://www.unicode.org/wg2/ <https://www.unicode.org/wg2/> where more
>     recent working documents are available,
>     and hopes are that ISO/IEC standards can be stabilized with normative
>     references
>     to Unicode specs like UTS Unicode Technical Standards and UAX Unicode Standard
>     Annexes.
> 
>      > + memmem is in string.h under the __GNU_VISIBLE guard. Does it now need
>     to be
>      > under a __POSIX_VISIBLE > 2024mmdd and __GNU_VISIBLE?
>      >
>      > + aligned_alloc in stdlib.h is only guarded by C > 2011. Does that need
>     to be an
>      > || POSIX >= 2024 also?
>      >
>      > + C11 and POSIX-2024 add these complex methods.
>      >
>      > double complex CMPLX(double x, double y);
>      > float complex CMPLXF(float x, float y);
>      > long double complex CMPLXL(long double x, long double y);
>      >
>      > FreeBSD has a macro implementation
>      > (https://github.com/lattera/freebsd/blob/master/include/complex.h#L49
>     <https://github.com/lattera/freebsd/blob/master/include/complex.h#L49>
>      > <https://github.com/lattera/freebsd/blob/master/include/complex.h#L49
>     <https://github.com/lattera/freebsd/blob/master/include/complex.h#L49>>). If I
>      > add that to complex.h, does it also need a __ISO_C_VISIBLE >= 2011 ||
>      > __POSIX_VISIBLE > 2024mmdd guard?
> 
>     The same minutes linked above says:
> 
>              "The ISO/IEC [POSIX] ballot closes on June 28."
> 
>     Also note that ISO/IEC 9899:2024 C is targeted for July 12, and JTC 1/SC 22/WG
>     14 hope to get that passed by some picky ISO editors in time, or it will be
>     cancelled for this year.
> 
>     So look for ISO C >= 2024 and add changes, including docs and compiler options,
>     and note 202X is becoming ambiguous so groups are using 202Y.
> 
>      > + Should I continue to look for guards on things that are newly added in
>     POSIX
>      > 2024 and already in newlib?
>      >
>      > That's it so far. Just picky stuff that I plan to address given some advice.
> 
>     I am also waiting to see if this will be issued as SUSv5?
> 
>     -- 
>     Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada
> 
>     La perfection est atteinte                   Perfection is achieved
>     non pas lorsqu'il n'y a plus rien à ajouter  not when there is no more to add
>     mais lorsqu'il n'y a plus rien à retirer     but when there is no more to cut
>                                       -- Antoine de Saint-Exupéry
> 

-- 
Take care. Thanks, Brian Inglis              Calgary, Alberta, Canada

La perfection est atteinte                   Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter  not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer     but when there is no more to cut
                                 -- Antoine de Saint-Exupéry