[Bug fortran/14959] New: [gfortran] (G95) not recognizing internal file in PURE/ELEMENATAL function

[Bug fortran/14959] New: [gfortran] (G95) not recognizing internal file in PURE/ELEMENATAL function

[jtk at ou dot edu]

The compiler is not recognizing the internal file when in a PURE/ELEMENTAL
function.  Removing PURE/ELEMENTAL removes the error, but breaks subsequent
code.



Code:

   ELEMENTAL FUNCTION FMT_GN(N)
!  "Gn" IO Format Specifier  
   USE CHAR_UTIL, ONLY: STRIP
   INTEGER(I4B),    INTENT(IN) :: N
   CHARACTER(LEN=12)           :: FMT_GN
!  Calling Sequence:
!  N       Width of Field
!  FMT_GN  Format character string
!
   INTEGER(I4B)     :: N2
   CHARACTER(LEN=4) :: C_NUM, C_NUM2
   INTEGER          :: IOS
!
   N2 = N - 7
   WRITE(C_NUM ,'(I4)',IOSTAT=IOS) N
   WRITE(C_NUM2,'(I4)',IOSTAT=IOS) N2
   FMT_GN = '(1PG'//STRIP(C_NUM)//'.'//STRIP(C_NUM2)//')'
!
   END FUNCTION FMT_GN



Error:

tmp/g95/bin/i686-pc-linux-gnu-g95 -g  -c FG_Text_File.f90
In file FG_Text_File.f90:874
                                                                                
   WRITE(C_NUM ,'(I4)',IOSTAT=IOS) N
                                   1
Error: io-unit in WRITE statement at (1) must be an internal file in a PURE
procedure
In file FG_Text_File.f90:875
                                                                                
   WRITE(C_NUM2,'(I4)',IOSTAT=IOS) N2
                                    1
Error: io-unit in WRITE statement at (1) must be an internal file in a PURE
procedure
make: *** [FG_Text_File.o] Error 1

-- 
           Summary: [gfortran] (G95) not recognizing internal file in
                    PURE/ELEMENATAL function
           Product: gcc
           Version: tree-ssa
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jtk at ou dot edu
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959

[Bug fortran/14959] [gfortran] not recognizing internal file in PURE/ELEMENATAL function

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-14 22:22 -------
Note g95 is slightly different than gfortran.

With gfortran I get:
Fatal Error: Can't open module file 'char_util.mod' for reading: No such file or directory

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[gfortran] (G95) not        |[gfortran] not recognizing
                   |recognizing internal file in|internal file in
                   |PURE/ELEMENATAL function    |PURE/ELEMENATAL function
   Target Milestone|---                         |tree-ssa


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959

[Bug fortran/14959] [gfortran] not recognizing internal file in PURE/ELEMENATAL function

[Tobias dot Schlueter at physik dot uni-muenchen dot de]

------- Additional Comments From Tobias dot Schlueter at physik dot uni-muenchen dot de  2004-04-30 14:06 -------
(In reply to comment #1)
> Note g95 is slightly different than gfortran.
> 
> With gfortran I get:
> Fatal Error: Can't open module file 'char_util.mod' for reading: No such file
or directory

This is because the testcase is not selfcontained.
   ELEMENTAL FUNCTION FMT_GN(N)
!  Gn IO Format Specifier  
   INTEGER(4),    INTENT(IN) :: N
   CHARACTER(LEN=12)           :: FMT_GN
!  Calling Sequence:
!  N       Width of Field
!  FMT_GN  Format character string
!
   INTEGER(4)     :: N2
   CHARACTER(LEN=4) :: C_NUM, C_NUM2
   INTEGER          :: IOS
!
   N2 = N - 7
   WRITE(C_NUM ,'(I4)',IOSTAT=IOS) N
   WRITE(C_NUM2,'(I4)',IOSTAT=IOS) N2
!
   FMT_GN = '(1PG'//C_NUM//'.'//C_NUM2//')'
   END FUNCTION FMT_GN
compiles flawlessly.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959

[Bug fortran/14959] [gfortran] not recognizing internal file in PURE/ELEMENATAL function

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-12 17:46 -------
Can you provide a self contained example?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
           Keywords|                            |rejects-valid


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959

[Bug fortran/14959] [gfortran] not recognizing internal file in PURE/ELEMENATAL function

[tobi at gcc dot gnu dot org]

------- Additional Comments From tobi at gcc dot gnu dot org  2004-05-12 23:01 -------
(In reply to comment #3)
> Can you provide a self contained example?
No, because I can't reproduce it. The testcase I gave removes the dependency on
the module, but it doesn't fail. Looking at the original bug submission again,
this seems to be a problem in g95, not gfortran.

J. Timothy, if your problem persists, and if I understand correctly that your
problem is in g95, I suggest you ask Andy Vaught about it (and suggest to him
that he works with the gfortran people again :-). According to their status
page, it's fixed there as well.

Independent of that, we would also appreciate if you tried gfortran on your
code, and tell us if it works.

I'm closing this bug as WORKSFORME. If you believe that you did actually find a
bug in gfortran, you're welcome to open it again. If you do so, please provide
more detailed information.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tobi at gcc dot gnu dot org
             Status|WAITING                     |RESOLVED
         Resolution|                            |WORKSFORME


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959

[Bug fortran/14959] [gfortran] not recognizing internal file in PURE/ELEMENATAL function

[jtk at octopus dot scif dot ou dot edu]

------- Additional Comments From jtk at octopus dot scif dot ou dot edu  2004-05-13 00:27 -------
Subject: Re:  [gfortran] not recognizing internal file in PURE/ELEMENATAL function

Hello,

The newest copy of g95 now makes it past this problem.  I now have other
problems that I will need to narrow down before submitting a bug report.
I do apologize for leaving in the dependency.  We use quite a few nested
modules, and many F90/95 features.  In fact, the sun compilers didn't
even generate anything except a divide by zero error compiling our code.
SGI and the Intel version 8 compilers also don't make it without
internal compiler errors.  Intel v7 and Compaq Digital Fortran do make
it.  All in all, it looks like the development is going quite nicely.

As a side note, what exactly is the difference between the gfortran and
g95 projects?  The home page for G95 points to the gfortran bugzilla
page.

Thanks,

Tim Kwiatkowski


On Wed, May 12, 2004 at 11:01:07PM +0000, tobi at gcc dot gnu dot org wrote:
> 
> ------- Additional Comments From tobi at gcc dot gnu dot org  2004-05-12 23:01 -------
> (In reply to comment #3)
> > Can you provide a self contained example?
> No, because I can't reproduce it. The testcase I gave removes the dependency on
> the module, but it doesn't fail. Looking at the original bug submission again,
> this seems to be a problem in g95, not gfortran.
> 
> J. Timothy, if your problem persists, and if I understand correctly that your
> problem is in g95, I suggest you ask Andy Vaught about it (and suggest to him
> that he works with the gfortran people again :-). According to their status
> page, it's fixed there as well.
> 
> Independent of that, we would also appreciate if you tried gfortran on your
> code, and tell us if it works.
> 
> I'm closing this bug as WORKSFORME. If you believe that you did actually find a
> bug in gfortran, you're welcome to open it again. If you do so, please provide
> more detailed information.
> 
> -- 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |tobi at gcc dot gnu dot org
>              Status|WAITING                     |RESOLVED
>          Resolution|                            |WORKSFORME
> 
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959
> 
> ------- You are receiving this mail because: -------
> You reported the bug, or are watching the reporter.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959

[Bug fortran/14959] not recognizing internal file in PURE/ELEMENATAL function

[Tobias dot Schlueter at physik dot uni-muenchen dot de]

------- Additional Comments From Tobias dot Schlueter at physik dot uni-muenchen dot de  2004-07-09 23:02 -------
Subject: Re:  [gfortran] not recognizing internal file
 in PURE/ELEMENATAL function


Sorry for replying late, looks like I missed that mail.

jtk at octopus dot scif dot ou dot edu wrote:
> As a side note, what exactly is the difference between the gfortran and
> g95 projects?  The home page for G95 points to the gfortran bugzilla
> page.

That page is not updated anymore, Doug Cox who had been doing that page
seems to have realized that it's not fair from g95 to obfuscate their
code and not submit back their fixes, but still use our bug database.

For information on the relationship between g95 and gfortran I point you
to this comp.lang.fortran thread:
http://tinyurl.com/2qbg8

- Tobi


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14959