http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48972
Summary: OPEN with Unicode file name
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Keywords: accepts-invalid, diagnostic
Severity: normal
Priority: P3
Component: fortran
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: bur...@gcc.gnu.org
CC: jvdeli...@gcc.gnu.org
This PR is motivated by the thread which started at
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=COMP-FORTRAN-90;59308f3c.1105
GNU Fortran happily accepts kind=4 character strings to the FILE= argument of
the OPEN statement - and probably also to the other string arguments.
However, the Fortran 2008 standard has:
R905 connect-spec is
...
or FILE = file-name-expr
with
R906 file-name-expr is scalar-default-char-expr
Thus, such strings should be rejected -- at least with -std=f2008.
* * *
Independent of that, it would be convenient if as vendor extension passing a
UCS-4 string would be allowed. The only problem is how it should be handled in
the library.
For Unix systems, I think converting the UCS-4 to UTF-8 and using it in the
normal file open should work.
However, for Windows, I think one needs a special solution as Windows seems to
use UTF-16 everywhere [1]. Thus, one should be able to directly pass the UCS-16
file name to CreateFileW [2].
[1] http://msdn.microsoft.com/en-us/library/dd374081%28v=vs.85%29.aspx
[2] http://msdn.microsoft.com/en-us/library/aa363858%28v=vs.85%29.aspx
Example program. Sample usage:
$ gfortran test.f90
$ ./a.out
Enter filename: ファイル
$
Should create "ファイル.dat" with the content "Hello World and Ni Hao -- 你好" - the
latter works but the file name is as written above "?" (= \343). If one passes
"44", the created file is just "4".
use iso_fortran_env
implicit none
integer, parameter :: ucs4 = selected_char_kind ('ISO_10646')
character(len=30, kind=ucs4) :: str
integer :: unit
open(unit=INPUT_UNIT, encoding='utf-8')
write(*, '(a)', advance='no') 'Enter filename: '
read(*,*) str
open(newunit=unit, file=trim(str)//ucs4_'.dat', encoding='utf-8')
write(unit, '(a)') ucs4_'Hello World and Ni Hao -- ' &
// char (int (z'4F60'), ucs4) &
// char (int (z'597D'), ucs4)
close(unit)
end
