To: J3 J3/19-206 Subject: DATA subgroup recommendations on feature list From: Malcolm Cohen Date: 2019-August-06 1. Introduction This paper describes the /DATA subgroup recommendations as to whether a suggested feature should be included in Fortran 202x. (In the first revision of this paper, only US15, US24, and US25 have recommendations.) US15. Virtuous (SIMPLE) procedures /DATA subgroup considers that this has beneficial effects both on optimisation and on program maintenance, and a small cost to the standard and on vendors. Therefore, /DATA recommends that US15 be included in F202x. US24. Generic programming /DATA subgroup notes that US24 mentions "containers"; however, it was decided long ago not to pursure container-specific features, but to pursue generic features that would also help producing containers. /DATA subgroup considers that we do not have enough time and resources to properly do Parameterised Modules (type-safe or not) without blowing both the schedule and performing the disservice to the vendor community of overloading them a new big feature while they are still catching up with previous editions of the standard. However, we do believe this feature is important enough to the user community that work should continue on investigation and development thereof. /DATA subgroup further considers that the same applies to Parameterised Procedures (aka Template Procedures), and that these also provide less functionality without a concomitant reduction in the compiler and standard burden. And that therefore these should not be proceeded with at this stage. /DATA subgroup considers that Intelligent Macros are sufficiently useful in handling enough of the simple key use cases for generics, useful in handling conditional compilation and rank genericity, and sufficiently self-contained not to be a great burden either on the standard or on the vendor. Furthermore, it is possible to write a stopgap preprocessor which would ease vendor adoption. Also, macros will still be useful if/when Parameterised Modules become available. Therefore the recommendation is: (1) Put Intelligent Macros into F202x. (2) Continue working on Parameterised Modules with a view to inclusion in a later standard, hopefully F202y. This should include some work by /DATA subgroup at future meetings, as time permits. US25. Exceptions /DATA subgroup considers that there is insufficient consensus, and more than sufficient complexity, to make it essentially impossible to get these into F202x. Therefore the recommendation is not to pursue these at this time. (Advocates of Exceptions are of course welcome to continue to develop their ideas and to seek consensus for a later standard.) ===END===