ISO/IEC JTC1/SC22/WG5-N2242

                 UK Activity Report, 2024-2025

The BSI (British Standards Institute) has asked the UK Fortran Group to 
operate in a more formal manner and now funds two on-line meetings 
with secretarial support per year. We decided that meeting shortly 
before each J3 or WG5/J3 meeting would be best and added a third meeting
using our own resources. 

Our first meeting took place on 19 September 2024 and is documented in 
J3/24-167. Almost all the discussion was on the list of potential new 
features in WG5-N2234. The proposal from Japan on generic subprograms, 
JP01, was strongly supported. We noted that the items varied widely in 
size and suggested the reuse of the 8-point scale in N1594 (2004) to 
provide a rough measure of the size of each feature. Details are 
in J3/24-167. J3 adopted the scale, see J3/24-185, pending further 
discussion of the individual levels and decided that a column showing 
the levels should be added to Standing Document 10, the F202y workplan.

Our second meeting took place on 16 January 2025 and is documented in 
J3/25-106. Almost all the discussion was on the list of potential new 
features. We expressed concern about the lack of progress on:
     US04 (level 7). Add asynchronous tasks.
and
     US10 (level 6). Add a Fortran-friendly preprocessor.
We thought that work on these would need to start as a matter of urgency 
if they are to be completed within the development schedule of the next
Standard.
 
Our third meeting took place on 8 May 2025. The discussions were:

1. We deeply regretted that the current J3 rules no longer allow 
joint meetings. Having the WG5 meeting after the J3 meeting worked 
very badly last year. The meeting was closed early on the first day, 
with no discussion of the work items. The meeting started early (one
of us joined at the original time, stated on the agenda) and the 
discussion of the work items was brief. The Panel thinks that it 
would be much better to have the WG5 meeting first so that strategic
decisions can be made before the detailed work is commenced by J3. 
The Chair was tasked with asking for this change to be made for the 
June meeting.  See response below.

2. There was general unease over the recent ‘secret’ ballot. The 
Panel felt that this is a break with tradition, obscures national 
positions, and is counter to the goal of reaching consensus.  
For consensus, all members need to see the votes of others and 
discuss their differences. The Chair was tasked with discussing
this with the Convener.  

3. The list of accepted work items was reviewed. The Panel was very
disappointed that Standing Document 10, the F202y workplan, has not been 
updated since October 2023. It used to be updated soon after each J3 
meeting. The Panel's view on progress remains that it has been too slow 
and therefore some reduction in features is needed. On the two items 
highlighted at our second meeting,

     US04 (level 7). Add asynchronous tasks. This should now be 
     removed from the work list. So much work is required that the
     next Standard is likely to be delayed. The functionality is 
     already available widely in extensions that support OpenMP. 

     US10 (level 6). Add a Fortran-friendly preprocessor. This 
     should be removed from the work list. While less work is 
     required than for US04, it could delay the next Standard.
     Since the purpose of standards is to facilitate the portability of 
     programs there is much less need for portability of preprocessors.  
     An earlier standard for preprocessors for Fortran 
     (ISO/IEC 1539-3:1999) was withdrawn for lack of use, given the 
     alternatives already available.  If there is real demand, the work 
     done on US10 could be used in a subsidiary standard.     
     
4. The Panel strongly disagreed with the Convener's decision (N2239) to 
put the three conditionally accepted items on the approved work list 
because all had NO votes and no attempt was made to reach consensus.
The panel then discussed whether it wanted the items to be on the 
approved work list:
     
     US20 (level 4)  Add SCAN and CO_SCAN. This should not be 
     added to the work list. There is very little demand for this from
     users. Equivalent features in other languages have been little 
     used. 

     US22/DIN2 (level 4) Add unions in interoperable derived types
     This should not be added to the work list. It is a form of 
     equivalence that is superseded by better forms of storage 
     management. 
     
     DIN1 (level 4) Add execution of collective procedures on a
     specified team. This is an obvious "hole" in the teams feature
     that should be filled.

ISO places great emphasis on consensus.  The word is used 69 times in 
the Directives, which include 'A working group operates by consensus' 
(1.12) and 'It is ISO's aim and expectation that each of its 
International Standards represents a worldwide consensus' (0.7f). 
Consensus has clearly not been reached for these three items.  For
US22/DIN2 for example the vote in the 2024 WG5 meeting was 7-5-8, and 
in the recent ballot was 20-5-1. They should be reconsidered.

The Chair sent an email to the Convener as tasked in items 1 and 2 
above. On item 1, he replied "At this point in the process, the idea is 
that J3, as the technical development body, has the most work to do 
and WG5’s official duties are to vote to add or remove items from the 
work list. Given that nearly everyone who attends WG5 meetings is also 
a J3 member, I don’t see an advantage to switching things around."
The panel disputes this.  Allocating the last three half-days of a 
meeting to WG5 is unsatisfactory.  The Friday afternoon session is 
seldom used in practice because members are tired and want to go home 
and the Friday morning is usually cut to a couple of hours at most.
There is therefore little time for serious technical discussion.

On item 2, the Convener promised not to use the ISO form in the future.