This is the mail archive of the libc-alpha@cygnus.com mailing list for the glibc project.


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

Re: glibc-2.0.93 homebrew distribution tests


John Kennedy <jk@csuchico.edu> writes:

>   Most portability problems were caused by people using their own
> prototypes rather than including the proper header files.  Prototypes for
> strdup() in particular, although string functions (strcmp(), strncmp(),
> strcpy() & strstr()) were popular.

This is a nono according to ISO C 89.  A program must always be
prepared to see the function being defined as macros.

>   Many configure programs laying around check for HAVE_UTMP_H and
> HAVE_UTMPX_H and end up in a strange state when you have both.  So far
> I've been able to fix all of them by ignoring HAVE_UTMPX_H if I have
> HAVE_UTMP_H defined.

utmpx.h is the standard interface.  utmp.h is there only for
compatibility reasons.

>   A bunch of programs expected "union semun" to be defined, but the
> lack of it was covered in comments found in the right header files
> (checking for _SEM_SEMUN_UNDEFINED, etc.).

I corrected this not long ago.  The Unix98 standard (and all of the
earlier standards of this line) make clear that the user has to define
this type.

>   e2fsprogs had a blkcnt_t defined of their own, which conflicted
> with 2.0.93's name.

blkcnt_t is a standard type and therefore must be defined in the
standard headers.

>   We used to have a <sys/module.h> at one point that prototyped
> get_kernel_syms().  I don't know where that is these days.

See the ChangeLog

        * sysdeps/unix/sysv/linux/sys/module.h: Remove file.  It's
        incompatible with Linux 2.1.x.  Packages using modules
        (e.g. modutils) should/do provide their own version.  Suggested by
        Richard Henderson.

Richard is maintaining the modutils package which now provides this
header.  It is simply not possible to provide one header in the libc
which provides the definitions since they very with the kernel.

>   Some things that filed to compile that seemed to cause compiler problems
> were lynx-2.8 & gnuplot-3.5 (croaking with sig11 & sig6, respectively).

I don't know the problem are, too, but if you can take a look with the
debugger this certainly would help.


For all the other problem: there is nothing which can be done in
glibc.  Consider contacting the authors so that the next versions can
get fixed.

Thanks,

-- Uli
---------------.      drepper at gnu.org  ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Cygnus Solutions `--' drepper at cygnus.com   `------------------------


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