Bug情況

最近寫Fortran代碼的時(shí)候遇到了個(gè)怪bug, 代碼如下:

然后離譜的來了, 我發(fā)現(xiàn)在gcc750和gcc710里面跑出來的shape_data的lbound都是0, 也就是說index是從0開始的, 像C一樣了...

但是在gcc11下面就沒問題, 是從1開始的

而且下面注釋那幾行打開的話也沒問題, 似乎就是傳參進(jìn)去才會(huì)有問題

module m1
  implicit none

contains
  subroutine m1s1(data)
    implicit none
    integer(kind=4),intent(in) :: data(:)

    integer, allocatable :: shape_data(:)

    allocate(shape_data, source=shape(data))
    ! allocate(shape_data(size(shape(data))), source=shape(data))   ! it's ok
    print *, ubound(shape_data), ubound(shape(data))
    print *, lbound(shape_data), lbound(shape(data))

  end subroutine
end module


program main
  use m1
  implicit none
  integer,allocatable :: a(:)


  call m1s1([1,2,3])
  
  !allocate(a, source=shape([4,4,5]))
  !print *, ubound(a)
  !print *, lbound(a)



end program

run一下:

DESKTOP-TKTNVE0:shape_bound$ gfortran --version
GNU Fortran (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0

DESKTOP-TKTNVE0:shape_bound$ gfortran main.F90
DESKTOP-TKTNVE0:shape_bound$ ./a.out
           0           1
           0           1

解決思路

• 要么換個(gè)編譯器
• 要么像上面注釋行一樣, 制定以下shape吧..但是這樣用source的意義小了...就是少寫一行吧那

總結(jié), 一個(gè)字, 離譜!