Project 31: Array processing in Fortran 8x

Madison 1982:
=============

Rice said that he accepted the decision to make no 
special provision for vectors and matrices but was unhappy 
about the absence of range variables, that operations on 
indices are difficult, that SUM is not sufficiently general, 
that WHERE is too complex and gives problems in conjunction 
with transformational functions, that IDENTIFY runs against 
the trend of changes being made since it is a kind of 
EQUIVALENCE.

Smith said that RANGE is felt to be unsafe since any 
ordinary statement may affect ranges. With the IDENTIFY 
changes are confined to the execution of this statement.

Rice mentioned that a meeting is planned in Washington 
for a national effort for a supercomputer a thousand times 
faster than any currently available. It is intended only 
for scientific use.

The following motion was proposed by Rice, seconded Lawson and was
approved by 10 votes to 1.

WG 2.5 believes in the importance of array features in Fortran 8x but
is unhappy with the totality of the present proposals and would like
to contribute to their refinement.

At a subcommittee meeting detailed discussions on the Fortran 8x array
facilities were begun. It was agreed that Smith, Rice and Reid would
prepare a paper for X3J3 (first draft: Reid) and would all try to
attend the November meeting of X3J3.

Pasadena 1984:
==============

Reid notes that although the full Fortran 8x language is unlikely to be 
in use until the 1990s, subsets of the array features are likely to be 
implemented as extensions to Fortran 77 (e.g. Cray have indicated their 
intention to do so). He then summarized the features and a long discussion
followed, see the minutes IFIP/WG 2.5 (Pasadena-24)1124, pp. 8-12.