forums.silverfrost.com

[wws]

In the following test, FTN95 seems to lose track of the fact that variable I is used as a ac-do-variable in the array constructor:

program init_bug
implicit none

integer :: i
character(11), parameter :: string="hello world"

! This compiles:
character, parameter :: up_string(len (string)) = &
(/ (string(i:i), i=1, len (string)) /)

! Yet this does not compile:
integer, parameter :: bytes(len (string)) = &
(/ (iachar (string(i:i)), i=1, len (string)) /)

print *, string
print *, up_string
print *, bytes

end program

FTN95 reports the following:

[FTN95/Win32 Ver. 5.10.0 Copyright (c) Silverfrost Ltd 1993-2007]
0012) integer, parameter :: bytes(len (string)) = &
0013) (/ (iachar (string(i:i)), i=1, len (string)) /)
*** I is not a PARAMETER, so cannot appear in an initialisation expression

1 ERROR [salford_init_bug1.F90] - Compilation failed.

Playing around with the example a bit, I tried making I a PARAMETER. (This should not be needed, and may be illegal.) Depending on case FTN95 either generated bad code or simply crashed!

[PaulLaidler]

Yes this looks like a bug.

For the moment you could remove the parameter attribute and do the assignment at via an executable statement at runtime.

[wws]

Thank you, Paul. Consider this a FYI, and not a critical need.

[PaulLaidler]

For a work-around you could probably use "/ignore 538" on the command line in this context.

[PaulLaidler]

This bug has now been fixed for the next release of FTN95.