Incorrect output on minloc and maxloc of empty array

[vzecca]

!Incorrect output: must be compiled and run
program intrinsic_mmloc_4
integer, allocatable, dimension(:) :: d
integer, allocatable, dimension(:,:) :: a
integer, dimension(2) :: b

allocate (d(0))
if (maxloc (d, 1) .ne. 0) print *,'maxloc fails'
allocate (a(1, 0))
b = minloc (a)
if (any (b .ne. 0)) print *,'minloc fails'
end program

[andrew_4619 · Accepted solution]

There are a number of topic on this area. see this one linked beklow and other linked topics maybe that will help. I didn't study your case specificaly.

https://community.intel.com/t5/Intel-Fortran-Compiler/Change-in-result-of-MINLOC-MAXLOC-functions-when-all-elements-of/m-p/1125542#M132793

https://community.intel.com/t5/Intel-Fortran-Compiler/Unhelpful-MINLOC-MAXLOC-documentation/m-p/1006947#M105091

[vzecca]

Thank you, Andrew, for the suggestion.

The correct execution of the code requires opetion -assume nooldmaxminloc

which adheres to the Fortran standard.

Unfortunately the default option violates the standard.

Closing this issue.

[Steve_Lionel]

You can use the option -standard-semantics to change defaults that don't match the standard. In all cases, these are situations where earlier versions of the standard were silent, and the developers chose an option that conflicted with a later revision. If existing code would break, or performance would suffer significantly, the default was left the same. 

See https://software.intel.com/content/www/us/en/develop/articles/intel-fortran-compiler-support-for-fortran-language-standards.html for a list of these conflicting defaults.