https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127733
Bug ID: 127733
Summary: Type parameter inquiry on an array complex part
designator (z%re%kind) is not scalar
Product: gcc
Version: 16.2.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: ivan.pribec at gmail dot com
Target Milestone: ---
A type parameter inquiry is expected to be scalar even if its designator is an
array (F2023 9.4.5, Note 1): "It is scalar even if *designator* is an array".
gfortran returns z%re%kind as a rank-1 array when z%re is an array complex part
designator, while z%kind is scalar.
program p
implicit none
complex :: z(3) = (1., 2.)
print '(a,t30,i2)', 'rank of z%kind', size(shape(z%kind))
print '(a,t30,i2)', 'rank of z%re%kind', size(shape(z%re%kind))
print '(a,t30,i2)', 'rank of z(2:3)%im%kind', size(shape(z(2:3)%im%kind))
print '(a,t30,*(i2))', 'z%re%kind', z%re%kind
end program
The output of this program is,
rank of z%kind 0
rank of z%re%kind 1
rank of z(2:3)%im%kind 1
z%re%kind 4 4 4
Because of the wrong rank, gfortran also rejects these valid statements
(z%re%kind is a constant expression):
integer, parameter :: kp = z%re%kind
Error: Array 'z' at (1) is a variable, which does not reduce to a
constant expression
k = z%re%kind ! integer :: k
Error: Incompatible ranks 0 and 1 in assignment at (1)
The same declarations with z%kind are accepted.
It looks like only complex-part designators are affected; inquiries on
derived-type components (non-complex) of arrays of structures (e.g. a%x%kind,
...) are scalar.
