------- Additional Comments From stevenb at suse dot de 2005-07-20 20:46
-------
Subject: Re: Null Characters instead of blanks in text output.
On Wednesday 20 July 2005 22:16, Steven Bosscher wrote:
> On Wednesday 20 July 2005 22:03, Steven Bosscher wrote:
> > On Wednesday 20 July 2005 21:07, Paul Thomas wrote:
> > > pinskia at gcc dot gnu dot org wrote:
> > > >------- Additional Comments From pinskia at gcc dot gnu dot org
> > > > 2005-07-20 15:41 ------- Confirmed, I think this was caused by:
> > > >2005-07-12 Paul Thomas <[EMAIL PROTECTED]>
> > >
> > > Guilty, as charged.
> > >
> > > Please find attached, ChangeLog entries, patch, the output that was
> > > incorrect and now looks the same as ifc, plus a testcase.
> >
> > This still doesn't fix mgrid either.
>
> FWIW I get the feeling that the sixtrack and mgrid problems are
> the same bug. For mgrid we have spurious '\0' characters in the
> output, and for sixtrack a read of an internal file with a format
> with an 'X' specifier in it ("READ(5,'(18X,I4)') NNO") fails...
> I still haven't been able to reduce this to a useful testcase :-(
Here is a test case that shows how fubar libgfortran is at the moment:
IMPLICIT NONE
INTEGER IOUT,IOA, III
REAL CC
INTEGER J(18),K(18)
CHARACTER C10*10
DO IOA=1,18
J(IOA) = IOA**2
END DO
OPEN(UNIT=18, FORM='FORMATTED', STATUS='UNKNOWN')
WRITE(18,'(I6,4X,18(2X,I4))') IOA,(J(III),III=1,18)
REWIND(18)
READ(18,'(A10,18(2X,I4))') C10, K
PRINT *,C10
DO IOA=1,18
PRINT *,K(IOA)
END DO
CLOSE(UNIT=18,STATUS='DELETE')
END
Output of libgfortran from 2005-07-09:
19
1
4
9
16
25
36
49
64
81
100
121
144
169
196
225
256
289
324
aka good.
Output of libgfortran from 2005-07-20 (today)
19
1
4
9
16
2
0
6
49
64
81
10
1
1
44
169
196
22
2
aka Very Wrong (tm).
Output of libgfortran from 2005-07-20 with the patches from
Jerry[1] and Paul[2] applied:
19
1
4
9
16
2
0
6
49
64
81
10
1
1
44
169
196
22
2
aka Still Very Wrong.
Guys, it seems we don't cover all the possibile output modes well for
libgfortran in our test suite :-(
Gr.
Steven
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22570
