Inconsistent warnings for conflicting initializations

[Harald1]

Hello,

the following non-conforming code is accepted by ifort/ifx, but gives a warning only with "ifort -stand" (not with ifx) and only for one variant:

program p
  implicit none
  type t1
     integer :: g = 0
  end type t1
  type t2
     integer :: g
  end type t2
  type(t1) :: u1 = t1(1)
  type(t2) :: u2 = t2(2) ! <- Warning only here, w/ ifort -stand
  data u1%g /3/ ! <- conflicting initialization
  data u2%g /4/ ! <- conflicting initialization
  print *, u1%g, u2%g
end program p

While NAG rejects both data statements (with slightly different error messages), Intel gives:

% ifx ifort-data2.f90 -stand
(silent = no warning at all)

% ifort ifort-data2.f90 -stand
ifort-data2.f90(10): warning #5436: Overlapping storage initializations encountered with U2
  type(t2) :: u2 = t2(2) ! <- Warning only here, w/ ifort -stand
----------------------^

While it is nice to have at least one warning here, it points to the wrong place: it is not the declaration that is at fault, it is the data statement.

[Barbara_P_Intel]

Thanks for reporting this. I filed a bug, COMPLRLLVM-50053. I'll let you know when a fix is available.

[Barbara_P_Intel · Accepted solution]

ifx 2024.1 now prints a warning message for this reproducer. This version was released last week. Please try it out!

[Harald1]

It's debatable which line to warn on - the declaration or the data statement - but now warnings are issued with -stand .

Thanks!