This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Documentation request for -ansi



(Regarding `-ansi' suppressing prototypes for erf/erfc/hypot,
 which makes them `int'):


On Thu, Jun 22, 2000 at 12:14:32AM +0200, Philipp Thomas wrote:

: > Repetitive trauma from -ansi

: IMO you've choosen the wrong list. This is a library issue, not one of gcc,
: so you should contact the glibc developers, not the gcc ones. Feel free to
: correct me if I've misunderstood you.


I sort of expected this answer; it's partially justified. But while I
can't positively identify the team RESPONSIBLE, that's not what I
want. Technically both GCC and libc behave correctly anyway.

Persons who use GCC/Glibc for numerical software will probably expect
one of two behaviours:

   o  Functions are undefined, and the user gets an error
      (could be in the linking phase ==> glibc issue)

   o  Compilation runs OK  ==>  Code works as expected

While these expectations are obvious, finding a technical solution
might not be, and might require cooperation/coordination between
gcc and libc.

I have been burnt twice; I'll probably remember the implications of
-ansi in future. However, the symptom seems to be more wide-spread
amongst numerical code, and others might fall into the same trap.
Perhaps that can be prevented.

I'll take it up with the glibc folks.

Regards,

Claus

-- 
claus.fischer@intel.com   Intel Corporation SC12-205 ... not speaking
phone   +1-408-765-6808   2200 Mission College Blvd.           for Intel
fax     +1-408-765-9322   Santa Clara, CA 95052-8119

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]