Error invoking datetime constructor with ifort-17.0.5 about datetime-fortran HOT 10 CLOSED

should be
utctime=datetime(year=2013,month=1,day=1)
but won't compile anyway

from datetime-fortran.

This compiler message usually implies that the calling signature in the constructor call does not match its interface. And because it doesn't, the compiler tries to default structure constructor, which doesn't work because components are declared private.

This program compiles OK for me:

program test

  use iso_fortran_env, only: real64
  use datetime_module, only: datetime, timedelta

  implicit none

  type(datetime) :: utctime

  utctime = datetime(2013, 1, 1, tz=-5._real64)
  utctime = datetime(year=2013, month=1, day=1)
  utctime = datetime(2013, 1, 1)

end program test

but tested with ifort 18:

$ ifort -V
Intel(R) Fortran Intel(R) 64 Compiler for applications running on Intel(R) 64, Version 18.0.0.128 Build 20170811
Copyright (C) 1985-2017 Intel Corporation.  All rights reserved.

I don't have v17 at hand so I can't test it right now.

Be extra careful when passing tz argument to the constructor, as of right now, it has to match the real64 kind size:

utctime = datetime(2013, 1, 1, tz=-5._real64)

from datetime-fortran.

made some progress after added
use iso_fortran_env, only: real64
and corrected tz
with Intel(R) Fortran Intel(R) 64 Compiler for applications running on Intel(R) 64, Version 18.0.0.128 Build 20170811
Still this error because of some mp stuff
test.o: In function MAIN__': test.f90:(.text+0x63): undefined reference to mod_datetime_mp_datetime_constructor_'

from datetime-fortran.

OK, it looks like all that is left is to link to datetime properly:

ifort test.f90 /path/to/datetime/libdatetime.a -ldatetime

from datetime-fortran.

I have that, it is some odd issue
ifort -c -L/apps/codes/datetime-fortran/build/lib -ldatetime -I/apps/codes/datetime-fortran/build/include test.f90

from datetime-fortran.

What does ifort say?

from datetime-fortran.

ifort 18 compiles fine except some problem with linking to the library that is being sorted out (may be related to mpi ifort on my computer)

ifort 17 fails like below

Intel(R) Fortran Intel(R) 64 Compiler for applications running on Intel(R) 64, Version 17.0.5.239 Build 20170817

PROGRAM test

USE iso_fortran_env, ONLY: real64
USE datetime_module, ONLY: datetime, timedelta

IMPLICIT NONE

TYPE(datetime) :: utctime

INTEGER :: i,j

utctime=datetime(year=2013,month=1,day=1,tz=-5._real64)
utctime=datetime(2013,1,1)

! utctime=utctime%utc()

END PROGRAM test

USE datetime_module, ONLY: datetime, timedelta
------^
test.f90(8): error #6406: Conflicting attributes or multiple declaration of name. [DATETIME]
TYPE(datetime) :: utctime
-------^
test.f90(12): error #6404: This name does not have a type, and must have an explicit type. [UTCTIME]
utctime=datetime(year=2013,month=1,day=1,tz=-5._real64)
--^
test.f90(12): error #6406: Conflicting attributes or multiple declaration of name. [DATETIME]
utctime=datetime(year=2013,month=1,day=1,tz=-5._real64)
----------^
test.f90(13): error #6406: Conflicting attributes or multiple declaration of name. [DATETIME]
utctime=datetime(2013,1,1)
----------^
test.f90(4): error #6580: Name in only-list does not exist. [DATETIME]
USE datetime_module, ONLY: datetime, timedelta
-----------------------------^
test.f90(4): error #6580: Name in only-list does not exist. [TIMEDELTA]
USE datetime_module, ONLY: datetime, timedelta
---------------------------------------^
compilation aborted for test.f90 (code 1)

from datetime-fortran.

I need to check it again fort ifort 17, will try again later and post errors

from datetime-fortran.

I figured the problem. intel 17 and 18 will both compile. If I can I would suggest one extension to the library in terms of generic functions: it would be beneficial to code datetime so it works with
datetime(year=2013,month=1,day=1,tz=-5._real64),
datetime(year=2013,month=1,day=1,tz=-5.), etc. i.e. tz as a real or an integer.
Thanks for your patience.

from datetime-fortran.

It would be straightforward to make the constructor generic to allow various kinds for tz argument. But this would increase the code size considerable for a feature with such small surface area. Unless there are more requests for it I'm not inclined to make it generic.

from datetime-fortran.