Re: glibc developers refuse to support user land fake FS syscalls

Hans Reiser (reiser@ceic.com)
Thu, 1 Jul 1999 22:56:51 +0000 (/etc/localtime)

Is glibc different from the kernel in this? Is it harder to modify the
glibc internals than kernel internals. Miklos, could you tell us how
you go about doing this technically, assuming I am a common C coder not
knowing much about glibc and libraries and other user space things? I
gather you don't just put a wrapper around write() that checks to see if
the pathname has your special syntax, and then call the real write()..?

Ulrich, this may be a problem for us, because there are some aspects of
FS functionality that are easier/better implemented in user space.

Hans

Ulrich Drepper writes:
> Hans Reiser <reiser@ceic.com> writes:
>
> > Stephen, I don't think the glibc folks are in the right about this one,
> > do you see any merit in their position? Is there anything I don't
> > understand about this that justifies the glibc developer's position?
>
> What you need to have such a library at user-level is access to many
> internals of the library. But this is not possible if one wants to
> have stability. For this reason the internals are not accessible. It
> is simply impossible to develop something if there is code which
> depends on the exact behaviour of certain internal functions. One
> cannot remove them, one cannot change them. Effectively this means no
> further development.
>
> Once there *is* no further development you can think about loosing
> these rules but even then it's too dangerious since you never can be
> sure there are no further changes required.
>
> --
> ---------------. drepper at gnu.org ,-. 1325 Chesapeake Terrace
> Ulrich Drepper \ ,-------------------' \ Sunnyvale, CA 94089 USA
> Cygnus Solutions `--' drepper at cygnus.com `------------------------

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/