Dan Nicholson wrote:
As in:
1. Do Ch.5
2. Do Ch.6, copy to different dir
3. Do Ch.6, diff with other dir
Or:
1. Do Ch.5
2. Do Ch.6, copy to different dir
3. Do Ch.5
4. Do Ch.6, diff with other dir
Hopefully the first way because the second way would produce exactly
the same thing and doesn't prove anything about the purity of the
build. It only shows the strength of Ch.5 in separating from the
host.
No. It shows the strength or weakness of the entire book, which is what
needs testing. This isn't just about chapter 6. The order of the general
packages (everything except the toolchain and testsuite packages) were
changed in both chapters 5 and 6. So both need testing.
This is what I've done:
1) Build *entire* LFS development version (minus the kernel) using
jhalfs and the LFS LiveCD as a host. Move the entire build to
LFS-20051110 on my partition.
2) Build *entire* LFS alphabetical version using jhalfs and the LFS
LiveCD as a host. Build a kernel, add some BLFS packages (libxml,
libxslt, subversion, wget), reboot into LFS alphabetical system.
3) Build *entire* LFS alphabetical using jhalfs and th built LFS
alphabetical system as a host.
This is what we're testing, the sanity of the entire book, not just the
results of chapter 6 using a known good toolchain.
--
JH
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
