Fortran Specialist Group Annual General Meeting 2015

1. Apologies for absence
2. Minutes of the previous AGM (2^nd October 2014)
3. Matters arising
4. Chairman's report
5. Treasurer's report
   SG Development Fund report
6. Membership Secretary's report
7. Web Editor's report
8. Election of Officers and Committee members
   (The following is taken from the Fortran SG Constitution:
   The name of any member accepting nomination for election or re-election as an Officer or as a Committee member shall be submitted in writing to the Secretary 7 clear days prior to the date of the AGM by two members of the Group with the written consent of the nominee.
   "In writing" applies equally to email or paper copy.)
9. Future activities for the Group
   
   
   • Joint meeting with IoP Computational Physics Group in 2016
   • Proposals for broadening the remit and appeal of the Group
10. Any other business
11. Date of next meeting

In February 2008, several coarray feature were removed from the draft standard with the understanding that once the new standard was completed, they would be the subject of a Technical Specification (TS). In 2011, WG5 decided that while the overall size of the addition should not be increased, the choice of features should be reviewed.

Significant changes were made up to the Feb. 2015 meeting of J3. A draft was submitted for ISO balloting in May 2015. This passed without any NO votes, but with 3 countries (Germany, Japan, and UK) voting YES with comments. All the comments were considered at the meeting of WG5 in London, August, 2015 and a modified TS was constructed, see document N2074 in N2074 in the WG5 documents page. A final check is being made and it is expected that very little changes will be needed before publication later this year.

The principal additional features involve teams, events, collectives, and support of continued execution in the presence of failed images. This talk will describe all the additional features. They will be included in Fortran 2015.

Fortran 2003 contains features for interoperability of Fortran with C, and they are widely implemented as extensions of Fortran 95 compilers. This provides for interoperability of procedures with non-optional arguments that are scalars, explicit-shape arrays, or assumed-size arrays, but not with arguments that are assumed-shape, allocatable, pointer, or optional.

To fill this gap, WG5 constructed a Technical Specification (ISO/IEC TS 29113:2012). This talk will describe the additional features. They will all be included in Fortran 2015.

The next revision of the Fortran standard, informally called Fortran 2015, incorporates the published corrigenda, the Technical Specifications on Further Operability with C and Additional Parallel Features and, apart from agreed editorial improvements, is otherwise restricted to the removal of simple deficiencies in, and discrepancies between, facilities in Fortran 2008.

A total of 34 such minor changes have been agreed and no more will be accepted, given that the standard is scheduled to be finally published in the summer of 2018. This talk gives an overview of the revisions.

There are now various coarray implementations available that support multi-image Fortran applications for a range of platforms. This talk will mention these implementations and move on to describe examples where coarrays have been used at scale.

At present the programmer has three options for handling character strings:

• CHARACTER type - introduced in Fortran 77 with new intrinsics in Fortran 90/95. They have limited flexibility and are risky because of the silent truncation on assignment and in other situations when length mismatches may occur, e.g. in procedure calls.
• ISO_VARYING_STRING - introduced with Fortran 90 in document ISO/IEC 1539-2:2000. It provides fully dynamic length for scalars and array-elements but they are easy to use only in expressions and assignments, not in I/O statements, nor in procedure calls (unless all code is converted to use the varying-string type).
• CHARACTER with allocatable length - introduced in Fortran 2003. The new allocatable-length string is very useful for scalar applications, but does not usefully extend to arrays.

These are at best only semi-compatible with each other.

In my opinion none of them provides all the facilities that are really desirable. I believe we need a fully-dynamic character string type, such that when you have an array of them, each element can have its own length. In particular length should be assigned automatically not only in assignment statements but in all other contexts in which the value of a variable is altered, e.g. in I/O statements and in procedure calls.