J3/05-283 To: J3 From: Michael Ingrassia Subject: Fortran Standard as Literature Date: 2005 November 10 I. Introduction The FORTRAN standard is written for and read by FORTRAN implementors. But it is also occasionally read "as literature" by technically proficient readers outside the relatively small community of Fortran implementors and Fortran standards mavens. [By this I refer to readers with no contingent claims to stake on the standard who merely wish to comprehend it.] First-time readers may draw incorrect technical conclusions if they are not able to discern aspects of the standard which are fully derivable from the standard text and part of the Fortran community consensus, but not written explicitly. Where the standard is not defective, it cannot and of course should not be changed by the Interpretation process. But the occasion of printing a new standard is an occasion for adding clarifying language about historically misunderstood aspects, if it can be done without in any way disturbing the existing consensus. Adding such language is a service to the wider technical community. II. Edit Add NOTE 7.16.5 at [130:4+] after "the processor may evaluate any mathematically equivalent expression, provided that the integrity of parentheses is not violated" : In particular, evaluating an expression using more precision than the nominal precision of the type of the expression is permitted. This does not change the type nor the size of the expression but may affect the value obtained. This latitude has been historically permitted to Fortran processors since the 1950's.