Bug of g77-3.0

Tim Prince tprince@computer.org
Mon Nov 12 01:46:00 GMT 2001


log() is determined by the library you use, not by gcc.  With glibc (usually
installed in linux), the version of log() will depend on the optimization
level you have chosen.  This would be a gcc bug only if some combination of
circumstances led to stack overflow, for example.  gcc-3.0 is obsolescent,
many people may not have the exact version available to test.  We would be
stabbing in the dark without more detailed information; at least the screen
output which you have when compiling 'g77 -v ....'
----- Original Message -----
From: "Stefano Simonucci,,," <stefano.simonucci@tin.it>
To: <bug-gcc@gnu.org>
Sent: Wednesday, November 21, 2001 12:55 AM
Subject: Bug of g77-3.0


> Hi
> I have a Debian (sid) distribution installed.
> I have successfully compiled (with g77-3.0) the following FORTRAN program:
>
>       double complex a,b
>       a=(-0.992198783,-0.0275850352)
>       write(6,*) ' A',a
>       b=log(a)
>       write(6,*) ' B',b
>       stop
>       end
>
> Nevertheless the program print the  "a"  variable but  not the  "b"
>  variable.
> I think that the log intrinsic function have a bug (maybe an infinite
> loop for particular
> values of the argument). The same function works in other cases.
> Thank you
> Good bye
>                   Stefano
>



More information about the Gcc-bugs mailing list