This is the mail archive of the 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: stropts.h


I did a 3-way diff between stropts.h in glibc, LiS and Solaris (2.6).  There
are quite a few differences between glibc and Solaris/LiS if you count
values of constants as differences.  There is more agreement between

Here is a minimalist suggestion:

Currently LiS is using _STR_OPTS_H as its inclusion marker.  I propose to
add "LIS_STROPTS_H".  I will change LiS stropts.h so that if STROPTS_H is
defined it will skip the vast majority of its content.  You would then
modify your stropts.h to test for LIS_STROPTS_H and not do anything if it is

This would at least keep the two files out of each other's way in case they
both get included.

However, because of the difference in the _values_ of constants, the user
will not get the expected behavior unless he/she includes LiS's stropts.h --
since those are the constants with which LiS itself was compiled.

Do you want me to just hack stropts.h into what I think it should be and
send it to you?

Do you guys have any actual library routines for getmsg, putmsg, etc?  Or do
you just have stubs that return EINVAL?  Maybe we should discuss providing
you with the actual LiS library routines that issue system calls on Linux.
If LiS isn't loaded the system calls return proper error codes from the

For the next version of LiS, I think that I will bring its constants more
into alignment with Solaris.  I will also check against UnixWare.

-- Dave

Ulrich Drepper wrote:

> David Grothe <> writes:
> > One of my LiS users wrote to report that he was experiencing conflicts
> > between the stropts.h in glibc and the stropts.h distributed with LiS.
> Shouldn't happen as I think I've copied correctly the prototypes and
> definitions from the standard.
> > I would like to know if you would be interested in some form of
> > coordination between these files.
> Since all which has to be done on our side is adjusting the headers if
> necesary I see no problems.
> --
> ---------------.      drepper at  ,-.   1325 Chesapeake Terrace
> Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
> Cygnus Solutions `--' drepper at   `------------------------

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