Στις Κυρ 04 Ιουν 2006 21:13, ο/η Jim Gifford έγραψε: > Hey everyone > The problem is that they will compile, but there is a lot of things > that also need to be fixed. If you look at the glibc-kern-header.sspec, > it basically does the same as the headers script. It's going to take a > long time before the headers get completely compliant with userspace > applications. In a previous email I showed information about a issue > that is the same on Sparc and MIPS, the page.h problem, but that doesn't > effect LFS, but it does effect CLFS.
Yep, In my case I meant they will compile *and* test ok (haven't built all of BLFS yet though). About your above example, it seems obvious that applications really *shouldn't* use stuff like that. Note I 'm only speaking from the LFS POV (don't have the hardware to do CLFS). I can understand busybox since it isn't supposed to be using glibc (normally), and maybe uclibc may not have the required workarounds yet, but these things should really be fixed in userspace! Moving the glibc hackery to support such things into kernel headers was *exactly* the thing linus will never accept into the kernel (and for good reason). OTOH, your script has worked very nice with well-behaved packages and keeping in mind that I 've always used it with different kernels than you tested it with, it seems pretty maintainable. Therefore, I still think your script together with git-hdrclean.patch is the way to go for the development version (or maybe an experimental branch) for the immediate future. (that's what I will use anyway) Just my 2c, Pantelis ____________________________________________________________________ http://www.freemail.gr - ������ �������� ������������ ������������. http://www.freemail.gr - free email service for the Greek-speaking. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
