IFIP WG 2.5 Project 82

Fortran Co-Arrays

Vienna 2014:
============

Sato suggested this new project at the 2013 meeting in Shanghai
but was unable to attend this years meeting. There was no report
on this project.

Amsterdam 2023:
===============

Update by John Reid received on 25-2-2023
-----------------------------------------
Coarrays are based on the Single-Program-Multiple-Data (SPMD) programming 
model, where the same executable program is replicated many times and 
executed independently. Each replication is known as an "image" and usually 
(but not necessarily) is executed on a separate process. Each image has its
own data objects, some of which are declared as coarrays. A coarray may be 
accessed from another image through an additional set of subscripts in 
square brackets, for example
       a(i,j)[k] = b
       c(:) = c(:)[p,q]
The programmer is responsible for ensuring that accesses to coarrays are
properly synchronized, for example, the statement
       sync all 
provides a barrier. 

The idea was proposed by Numrich in 1994 and formalized by Numrich and 
Reid in 1998 (Co-Array Fortran for parallel programming, ACM Fortran 
Forum, 17, 2). It was included in Fortran 2008 and the idea was extended
in Fortran 2018. For a full description, see chapters 17 and 20 of 
Metcalf, Reid, and Cohen (2018) (Modern Fortran Explained, second edition  
Oxford University Press). For the history of its inclusion in Fortran, 
see Reid, Long, and Steidel (2020) History of Coarrays and SPMD 
Parallelism in Fortran (Proc. ACM Program. Lang. 4, HOPL, Article 72).
-----------------------------------------


It was decided to close this project. Any sequel to this project can be found under Project 20.