Re: (impl) glibc patches
On Friday 30 August 2002 11:22, Wichmann, Mats D wrote:
> > On Friday 30 August 2002 10:57, Wichmann, Mats D wrote:
> > > - patches for "nice", which are not glibc cvs, and
> > > were also controversial with Debian for compatibility
> > > reasons
> > care to explain this one better?
> This is the stuff designed to make nice follow the
> SUS spec:
> RETURN VALUE
> Upon successful completion, nice() returns the new nice value
> minus [NZERO]. Otherwise, -1 is returned, the process' nice
> value is not changed, and errno is set to indicate the error.
> This is not the same as traditional Linux nice has done,
> where 0 indicates success and -1 failure.
> Someone from Debian has objected to this change being
> required in case something depends on the old behavior.
> The patch is described on the LSB Testing FAQ page:
> as fixes for T.nice 7,8. That patch is incomplete;
> the other piece of it is that a stub
> sysdeps/unix/sysv/linux/nice.c must be added which
> contains only "#include <sysdeps/unix/nice.c>"
> so that the glibc build system will pick it up
> in a build-for-Linux.
Well is the LSB going to force distribution change? Do we want to run the
risk of breaking existing code?
This sounds like a policy decision.