[J3] Upcoming gfortran 15 will contain an implementation of unsigned numbers

[Jeff Hammond]

I strongly encourage gfortran to issue a warning when this option is used that the code is not portable.

I have spent the past year removing gfortran-isms from HPC codes, in order to get them to compile with flang-new.  I am tired of Fortran users deciding that “compiles with gfortran” = “standard compliant”.  It hurts our ecosystem and undermines the purpose of WG5 when we encourage such behavior.

Jeff

From: J3 <j3-bounces at mailman.j3-fortran.org> on behalf of Brad Richardson via J3 <j3 at mailman.j3-fortran.org>
Date: Monday, 14. October 2024 at 17.54
To: j3 at mailman.j3-fortran.org <j3 at mailman.j3-fortran.org>
Cc: Brad Richardson <everythingfunctional at protonmail.com>
Subject: Re: [J3] Upcoming gfortran 15 will contain an implementation of unsigned numbers
External email: Use caution opening links or attachments


Hi Thomas,

I appreciate the gfortran developers experimenting and prototyping the
features they are interested. However, you may want to be aware that
the specific proposal (24-116), was not approved for the F202Y worklist
(https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwg5-fortran.org%2FN2201-N2250%2FN2234.txt&data=05%7C02%7Cjehammond%40nvidia.com%7C804e2e8493454c18652e08dcec602389%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C638645144889472870%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=cRhaxSNCO6cge0BSzczQgMDV4%2Fp2BRpWQuXO1LzxpFI%3D&reserved=0<https://wg5-fortran.org/N2201-N2250/N2234.txt>). At any rate, it will
be interesting to see how this works out.

Regards,
Brad

On Sun, 2024-10-13 at 13:08 +0200, Thomas König via J3 wrote:
> Hello,
>
> you may be interested to know that the upcoming release of Gfortran
> 15
> (due in early 2025) will include an experimental implementation of
> 24-116.txt, the unsigned number proposal. This is already part of the
> current development version, and is selected via the -funsigned
> option.
> The only major part missing at the moment is ISO_Fortran_binding.h .
>
> Some preliminary documentation can be found at
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgcc.gnu.org%2Fonlinedocs%2Fgfortran%2FUnsigned-integers.html&data=05%7C02%7Cjehammond%40nvidia.com%7C804e2e8493454c18652e08dcec602389%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C638645144889492223%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=4NVkMQcMN2%2FgF7EX7pGOJcowbP%2FJ6j3QpuzKDVrGPdw%3D&reserved=0<https://gcc.gnu.org/onlinedocs/gfortran/Unsigned-integers.html> ,
> this will be extended before the release.
>
> Test cases, which can also serve as example use cases, can be found
> at
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgcc.gnu.org%2Fgit%2F%3Fp%3Dgcc.git%3Ba%3Dtree%3Bf%3Dgcc%2Ftestsuite%2Fgfortran.dg&data=05%7C02%7Cjehammond%40nvidia.com%7C804e2e8493454c18652e08dcec602389%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C638645144889506152%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=vwewnzDRsuxarWHikh3312%2BJWzYmifvDS%2BBZTKVjvTU%3D&reserved=0<https://gcc.gnu.org/git/?p=gcc.git;a=tree;f=gcc/testsuite/gfortran.dg>
> (any file with "unsigned" in the name).
>
> Anybody who feels like compiling the current gcc development version
> can
> try this out already. Bug reports and comments welcome!
>
> Best regards
>
>         Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20241014/77fb672c/attachment.htm>