On Sun, Aug 05, 2001 at 09:58:03PM -0700, Daniel Jacobowitz wrote: > On Mon, Aug 06, 2001 at 12:53:29AM -0400, Ben Collins wrote: > > > The glibc build rules doesn't correctly find the headers installed. > > > There are bugs filed on this, both on glibc and kernel-headers (for > > > providing too many) but I can't find the numbers because brainfood > > > is down. > > > > The glibc rules aren't responsible for "finding" anything. They are > > responsible for defining them. If ppc doesn't have 2.4 kernel headers, > > there's the bug, not in glibc. > > As I told you on IRC, I don't think there's anything wrong with > depending on kernel-header-2.4 - IF that provide evaluates to something > sensible. It doesn't. I can just force it by keeping kernel-headers > for PPC installed on the build daemon; I'll do that after my next fight > with bitkeeper. But I strongly dislike buildd-operator-voodoo hacks to > make dependencies work.
You could select one of the alternatives that provide kernel-headers-2.4 in /etc/sbuild.conf, just like dozens of others are selected. It would be insane for me to try and keep up a list of exact kernel headers for each and every arch (there are > 10 now, I think), since things get outdated so quickly, and the build-dep becomes useless (just look at potato's kernel-headers build-deps). So instead, I opt for the generic build-dep, and let each arch resolve it appropriately. The alternative would be an even bigger mess. Ben -- .----------=======-=-======-=========-----------=====------------=-=-----. / Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \ ` [EMAIL PROTECTED] -- [EMAIL PROTECTED] -- [EMAIL PROTECTED] ' `---=========------=======-------------=-=-----=-===-======-------=--=---'
