Bug in gfortran when reading end of file?
in [Fortran]
|
Prev: Naive implementation of parallel Mersenne Twister (MT) pseudorandom number generator
Next: Naive implementation of parallel Mersenne Twister (MT) pseudorandom number generator
From: angelv on 24 Apr 2008 11:05 Hi all, I have found the following problem with gfortran. A minimal code: PROGRAM TEST IMPLICIT NONE INTEGER :: n, i, dat READ*, n DO i = 1,n READ*, dat PRINT*, dat END DO END PROGRAM TEST And an input file (note that the file does not end in a newline): [angelv(a)arce ~/temp]$ cat test.input 3 45 567 8945[angelv(a)arce ~/temp]$ Now, Intel Fortran and PGI have no problems with that, but gfortran reports and End of file, as it can be seen below. What "should" be the correct behaviour? Thought of asking here before reporting it as a bug to gfortran. Thanks, Ángel [angelv(a)arce ~/temp]$ ifort -o test test.f90 [angelv(a)arce ~/temp]$ ./test < test.input 45 567 8945 [angelv(a)arce ~/temp]$ pgf95 -o test test.f90 [angelv(a)arce ~/temp]$ ./test < test.input 45 567 8945 [angelv(a)arce ~/temp]$ gfortran -o test test.f90 [angelv(a)arce ~/temp]$ ./test < test.input 45 567 At line 8 of file test.f90 Fortran runtime error: End of file [angelv(a)arce ~/temp]$
From: utab on 24 Apr 2008 12:59 On Apr 24, 5:05 pm, angelv <txibi...(a)gmail.com> wrote: > Hi all, > > I have found the following problem with gfortran. A minimal code: > > PROGRAM TEST > IMPLICIT NONE > > INTEGER :: n, i, dat > > READ*, n > DO i = 1,n > READ*, dat > PRINT*, dat > END DO > END PROGRAM TEST > > And an input file (note that the file does not end in a newline): > > [angelv(a)arce ~/temp]$ cat test.input > 3 > 45 > 567 > 8945[angelv(a)arce ~/temp]$ > > Now, Intel Fortran and PGI have no problems with that, but gfortran > reports and End of file, as it can be seen below. What "should" be the > correct behaviour? Thought of asking here before reporting it as a bug > to gfortran. > > Thanks, > Ángel > > [angelv(a)arce ~/temp]$ ifort -o test test.f90 > [angelv(a)arce ~/temp]$ ./test < test.input > 45 > 567 > 8945 > [angelv(a)arce ~/temp]$ pgf95 -o test test.f90 > [angelv(a)arce ~/temp]$ ./test < test.input > 45 > 567 > 8945 > [angelv(a)arce ~/temp]$ gfortran -o test test.f90 > [angelv(a)arce ~/temp]$ ./test < test.input > 45 > 567 > At line 8 of file test.f90 > Fortran runtime error: End of file > [angelv(a)arce ~/temp]$ Hey, Are you sure of that, on my machine it seems perfectly OK 06:56 PM utab(a)PMA-05-013 ~ $ gfortran for.f90 06:56 PM utab(a)PMA-05-013 ~ $ vi dat 06:57 PM utab(a)PMA-05-013 ~ $ ./a.out <dat 45 567 8945 Rgds,
From: utab on 24 Apr 2008 13:05 On Apr 24, 6:59 pm, utab <umut.ta...(a)gmail.com> wrote: > On Apr 24, 5:05 pm, angelv <txibi...(a)gmail.com> wrote: > > > > > Hi all, > > > I have found the following problem with gfortran. A minimal code: > > > PROGRAM TEST > > IMPLICIT NONE > > > INTEGER :: n, i, dat > > > READ*, n > > DO i = 1,n > > READ*, dat > > PRINT*, dat > > END DO > > END PROGRAM TEST > > > And an input file (note that the file does not end in a newline): > > > [angelv(a)arce ~/temp]$ cat test.input > > 3 > > 45 > > 567 > > 8945[angelv(a)arce ~/temp]$ > > > Now, Intel Fortran and PGI have no problems with that, but gfortran > > reports and End of file, as it can be seen below. What "should" be the > > correct behaviour? Thought of asking here before reporting it as a bug > > to gfortran. > > > Thanks, > > Ángel > > > [angelv(a)arce ~/temp]$ ifort -o test test.f90 > > [angelv(a)arce ~/temp]$ ./test < test.input > > 45 > > 567 > > 8945 > > [angelv(a)arce ~/temp]$ pgf95 -o test test.f90 > > [angelv(a)arce ~/temp]$ ./test < test.input > > 45 > > 567 > > 8945 > > [angelv(a)arce ~/temp]$ gfortran -o test test.f90 > > [angelv(a)arce ~/temp]$ ./test < test.input > > 45 > > 567 > > At line 8 of file test.f90 > > Fortran runtime error: End of file > > [angelv(a)arce ~/temp]$ > > Hey, > > Are you sure of that, on my machine it seems perfectly OK > > 06:56 PM utab(a)PMA-05-013 ~ $ gfortran for.f90 > 06:56 PM utab(a)PMA-05-013 ~ $ vi dat > 06:57 PM utab(a)PMA-05-013 ~ $ ./a.out <dat > 45 > 567 > 8945 > > Rgds, And my version is GNU Fortran 95 (GCC) 4.1.2 20061115 (prerelease) (Debian 4.1.1-21)
From: glen herrmannsfeldt on 24 Apr 2008 13:31 angelv wrote: (snip) > And an input file (note that the file does not end in a newline): Your input file is defective, so that either answer is right. -- glen
From: Ron Shepard on 24 Apr 2008 13:30
In article <9db8499d-d05b-4ca5-9b6b-bbf8422ba231(a)59g2000hsb.googlegroups.com>, angelv <txibilis(a)gmail.com> wrote: > Now, Intel Fortran and PGI have no problems with that, but gfortran > reports and End of file, as it can be seen below. What "should" be the > correct behaviour? It is up to the compiler to handle it the way it wants. Many people prefer the way gfortran works in this case, and they think that the other way is inconvenient because it does not identify the incorrectly terminated record in the input file. $.02 -Ron Shepard |