This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: libio in gcc in CVS is not compatible with glibc 2.2
- To: gcc-bugs at gcc dot gnu dot org
- Subject: Re: libio in gcc in CVS is not compatible with glibc 2.2
- From: "John David Anglin" <dave at hiauly1 dot hia dot nrc dot ca>
- Date: Wed, 19 Jul 2000 00:23:34 -0400 (EDT)
- Cc: aoliva at redhat dot com, drepper at cygnus dot com
> On Tue, Jul 18, 2000 at 04:11:37PM -0300, Alexandre Oliva wrote:
>> On Jul 18, 2000, Ulrich Drepper <drepper@redhat.com> wrote:
>>
>> > Alexandre Oliva <aoliva@redhat.com> writes:
>> >> Then why does the test fail? Is it just a compiler error? I had
>> >> understood it was a run-time error, but maybe I just assumed too much.
>>
>> > It's a compile-time problem.
>>
>> Ok, so what can GCC for people to be able to build it with glibc 2.2
>> as well as older versions of glibc? Wouldn't changing the declaration
>> and the definitio of _IO_seekoff in GCC's libio break builds with
>> older releases of glibc?
>>
>
>Please note that I believe _IO_off_t is the same as _IO_fpos_t on all
>but glibc 2.2. I think it is safe. But someone has to test it on
>other platforms.
I believe the breakage is not limited to just glibc 2.2 platforms
_IO_off_t and _IO_fpos_t differ under VAX Ultrix. See my report
<http://gcc.gnu.org/ml/gcc-bugs/2000-06/msg00541.html> for details.
There are also many inconsistencies between the header prototype definitions
and the implementations resulting from the switch to _IO_off_t that need
to be corrected. These show up immediately when _IO_off_t and _IO_fpos_t
differ.
Dave
--
J. David Anglin dave.anglin@nrc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6605)