[Bug fortran/31199] New: wrong code generated with gfortran

[Bug fortran/31199] New: wrong code generated with gfortran

[jv244 at cam dot ac dot uk]

With a recent gfortran, the following compiles, but generates the wrong
results:

       program write_write
       character(len=20) :: a,b,c
       write (10,"(a,t1,a,a)") "XXXXXXXXX", "ABC", "DEF"
       write (10,"(a,t1,a)",advance='no') "XXXXXXXXX", "ABC"
       write (10,"(a)") "DEF"
       write (10,"(a)") "ABCDEFXXX"
       REWIND(10)
       read(10,*) a
       read(10,*) b
       read(10,*) c
       IF (a.NE.b) CALL ABORT()
       IF (a.NE.c) CALL ABORT()
       end


-- 
           Summary: wrong code generated with gfortran
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jv244 at cam dot ac dot uk


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[burnus at gcc dot gnu dot org]

------- Comment #1 from burnus at gcc dot gnu dot org  2007-03-16 13:54 -------
Current result:
 a = ABCDEFXXX
 b = ABCDEF
 c = ABCDEFXXX
Expected (g95/ifort):
 ABCDEFXXX
 ABCDEFXXX
 ABCDEFXXX
(NAG f95 has:
 ABCDEFXXX
 ABCXXXXXXDEF
 ABCDEFXXX
)


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jvdelisle at gcc dot gnu dot
                   |                            |org, burnus at gcc dot gnu
                   |                            |dot org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Keywords|                            |wrong-code
   Last reconfirmed|0000-00-00 00:00:00         |2007-03-16 13:54:17
               date|                            |
            Summary|wrong code generated with   |write with "t1" format gives
                   |gfortran                    |wrong output


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #2 from jvdelisle at gcc dot gnu dot org  2007-03-16 22:57 -------
I will work this one.  I want to review the standard.  I think we are
truncating and I am not so sure this has to do with the t1 format specifier. 
We'll see.  


-- 

jvdelisle at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jvdelisle at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2007-03-16 13:54:17         |2007-03-16 22:57:08
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #3 from jvdelisle at gcc dot gnu dot org  2007-03-17 15:59 -------
*** Bug 31207 has been marked as a duplicate of this bug. ***


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #4 from jvdelisle at gcc dot gnu dot org  2007-03-19 05:49 -------
Found it.  In the case of advance="no" we are not saving the maximum position
reached to be used by the following write statement.  This one is subtle.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[burnus at gcc dot gnu dot org]

------- Comment #5 from burnus at gcc dot gnu dot org  2007-03-19 11:12 -------
Current result:
 a = ABCDEFXXX
 b = ABCDEF
 c = ABCDEFXXX

Result by g95/ifort:
 ABCDEFXXX
 ABCDEFXXX
 ABCDEFXXX

Result by NAG f95, SUN and HP:
 ABCDEFXXX
 ABCXXXXXXDEF
 ABCDEFXXX

I think the latter is correct:

Section 9.2.3.3 of the Fortran 2003 standard:
"For a formatted stream output statement, ... the terminal point of
the file is set to the highest-numbered position to which data was
transferred by the statement."

(Thanks to Ian Hounam/NAG for pointing out.)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[dominiq at lps dot ens dot fr]

------- Comment #6 from dominiq at lps dot ens dot fr  2007-03-19 12:15 -------
> Result by g95/ifort:

You can probably add xlf to this list (should be checked
on a recent version).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" format gives wrong output

[dominiq at lps dot ens dot fr]

------- Comment #7 from dominiq at lps dot ens dot fr  2007-03-19 12:18 -------
> You can probably add xlf to this list

and Portland Group Fortran.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" + nonadvancing transfer format gives wrong output

[burnus at gcc dot gnu dot org]

------- Comment #8 from burnus at gcc dot gnu dot org  2007-03-19 12:53 -------
Reading further, I find:
"For nonadvancing input [...] If no error condition occurred in a nonadvancing
output statement, the file position is not changed."

If I understand the whole correctly, it means:
- The position for the nonadvancing transfer of "XXXXXXXXX", t1, "ABC" is after
the "ABC".
- After doing the advancing transfer "DEF" it is at the highest-numbered
position to which data was transferred by the statement, namely after
"ABCDEFXXX" + new_line.

Thus, NAG f95/Sun/HP seems to be wrong; ifort, g95, xlf and portland seem to be
right and gfortran gets it differently wrong.


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|write with "t1" format gives|write with "t1" +
                   |wrong output                |nonadvancing transfer format
                   |                            |gives wrong output


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" + nonadvancing transfer format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #9 from jvdelisle at gcc dot gnu dot org  2007-03-20 00:41 -------
Reply to comment #8: that was my take on this, that ifort is correct.  It
passed my intuitive sense of it.  Non-advancing means what it says so that the
subsequent write must continue from there and overwrite what was previously
written.  Then,  bytes written already after that are still there and define
the end of the record/line.

Regardless, I am close to a fix.  I will post to list in the next day or so. 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" + nonadvancing transfer format gives wrong output

[patchapp at dberlin dot org]

------- Comment #10 from patchapp at dberlin dot org  2007-03-23 16:20 -------
Subject: Bug number PR31199

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2007-03/msg01554.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" + nonadvancing transfer format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #11 from jvdelisle at gcc dot gnu dot org  2007-03-26 04:23 -------
Subject: Bug 31199

Author: jvdelisle
Date: Mon Mar 26 04:23:15 2007
New Revision: 123205

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=123205
Log:
2007-03-25  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR libgfortran/31199
        *io/io.h: Add saved_pos to gfc_unit structure.
        *io/open.c (new_unit): Initialize saved_pos.
        *io/transfer.c (data_transfer_init): Set max_pos to value in saved_pos.
        (next_record_w): Fix whitespace.
        (finalze_transfer): Calculate max_pos for ADVANCE="no" and save it for
        later use.  If not ADVANCE="no" set saved_pos to zero.

Modified:
    trunk/libgfortran/ChangeLog
    trunk/libgfortran/io/io.h
    trunk/libgfortran/io/open.c
    trunk/libgfortran/io/transfer.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" + nonadvancing transfer format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #12 from jvdelisle at gcc dot gnu dot org  2007-03-26 04:41 -------
Subject: Bug 31199

Author: jvdelisle
Date: Mon Mar 26 04:41:29 2007
New Revision: 123207

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=123207
Log:
2007-03-25  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR libgfortran/31199
        * gfortran.dg/fmt_t_4.f90: New test.

Added:
    trunk/gcc/testsuite/gfortran.dg/fmt_t_4.f90
Modified:
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199

[Bug fortran/31199] write with "t1" + nonadvancing transfer format gives wrong output

[jvdelisle at gcc dot gnu dot org]

------- Comment #13 from jvdelisle at gcc dot gnu dot org  2007-03-28 01:48 -------
Not a regression, closing.


-- 

jvdelisle at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31199