Matthew Burgess wrote:
Jim Gifford wrote:
This means userspace still will be broken until we submit patches to
the upstream maintainers for the programs that are broke.
And that's exactly what my proposal suggested we tackle. The way I
see it, if a package fails to build using the headers_install kernel
headers it's because:
1) headers_install didn't install a header that provides a public API or
2) the package is (ab)using an internal kernel header, that by pure
historical accident happened to be available until recently.
In the case of 1) we need to submit patches to the kernel to export
the API to userspace. In the case of 2) we need to submit patches to
the userspace package that makes it use a public API instead of using
the internal kernel header(s).
Regards.
Matt.
Matt, I agree, but, what I think is wrong is the idealism that David
Woodhouse has on this, he doesn't care if something breaks. A really
good example here is iptables. How would you suggest handling that can
of worms.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
