Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Bug in egcs 1.1.2

0 views
Skip to first unread message

JA

unread,
Jan 20, 2000, 3:00:00 AM1/20/00
to
In the one of the latest releases of the Fortran
compiler included into the egcs1.1.2 package
there is a bug in file operations. The users,
including me, of the Monte Carlo program MCNP
complain about that. The extracted pure form
of this bug can be examined with a short
program written by Jeff Siebers.

g77 up to 0.5.23 seem to be free of this bug.

Is any chance that a work around and/or a correction
can be implemented?

This bug has been reported to EGCS people with no
response.

Jacek Arkuszewski, Switzerland

comp.f

"Jochen Küpper"

unread,
Jan 21, 2000, 3:00:00 AM1/21/00
to
JA <lub...@datacomm.ch> writes:

> [1 <text/plain; us-ascii (7bit)>]

They have been pretty responsive for me.
You need to use the latest version, though. That is gcc-2.95.2,
IIRC. Or probably that + the latest snapshot as well.
If these still show the bug and you can cut the code down to a
reasonable piece of code, I am sure you will get a response soon.
Wether they can fix it out of the box is something different :-)

-- Jochen
Einigkeit und Recht und Freiheit
Libert`e, `Egalit`e, Fraternit`e
Sex, drugs and rock'n'roll

JA

unread,
Jan 21, 2000, 3:00:00 AM1/21/00
to

Jochen,

Forgot to mention that this bug is still in egcs-2.95 as of
19990628 pre-release. According to Skip Egdorf <h...@lanl.gov>
the use of endfile destroys the readability of the file.

Up to gcc-27.2.3 and g77-0.5.21 (perhaps even up to g77-0.5.23)
everything was OK.

Jacek Arkuszewski

Mumit Khan

unread,
Jan 26, 2000, 3:00:00 AM1/26/00
to
In article <388790D9...@datacomm.ch>, JA <lub...@datacomm.ch> wrote:
>
>In the one of the latest releases of the Fortran
>compiler included into the egcs1.1.2 package
>there is a bug in file operations. The users,
>including me, of the Monte Carlo program MCNP
>complain about that. The extracted pure form
>of this bug can be examined with a short
>program written by Jeff Siebers.
>
>g77 up to 0.5.23 seem to be free of this bug.
>
>Is any chance that a work around and/or a correction
>can be implemented?

Jeff had indeed submitted a bug report, and Toon (g77 guru and a regular
here) has submitted a fix for this bug, and will show in the next release.
You can of course rebuild your current release with his patch and hope for
the best. See:

Linkname: Patch [Fortran]: Fix endfile handling (libf2c)
URL: http://gcc.gnu.org/ml/gcc-patches/2000-01/msg00890.html

Regards,
Mumit

Toon Moene

unread,
Jan 26, 2000, 3:00:00 AM1/26/00
to
kh...@xraylith.wisc.edu (Mumit Khan) wrote:

> In article <388790D9...@datacomm.ch>, JA <lub...@datacomm.ch> wrote:

[ ... program using write/backspace/endfile/rewind/read sequence aborts ... ]

> >g77 up to 0.5.23 seem to be free of this bug.

> Jeff had indeed submitted a bug report, and Toon (g77 guru and a regular


> here) has submitted a fix for this bug, and will show in the next release.

It's correct that I submitted the fix, but - to be brutally honest - David M.
Gay wrote it :-)

--
Toon Moene (mailto:to...@moene.indiv.nluug.nl)
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
Phone: +31 346 214290; Fax: +31 346 214286
GNU Fortran: http://gcc.gnu.org/onlinedocs/g77_news.html

0 new messages