Hi everyone, I've done an ICA/Farce report on the LFS Alphabetical branch. The build was done on anduin despite what seem to be some hardware issues. Hopefully that hasn't affected the binaries. I don't have time to do much with the results right now, but I thought I'd put them up for public consumption before I leave town.
First, let me get some caveats about the building out of the way. I
used my own scripts, attempting to be as faithful to the book as
possible. C{,XX}FLAGS and LDFLAGS were unset throughout the build.
Also, I changed the Ch. 5 binutils-pass2 build slightly to allow for
easier restarts of the chroot. It should be using the exact some
files, though. For the ICA, I only did Ch. 5 and Ch. 6. This means
the bootscripts, startup files and kernel were not built. Next time,
I'll leave the kernel in, but the startup files were left out to
assure a consistent environment.
To see the scripts in their entirety, you can download them from
http://students.washington.edu/dbnichol/lfs/dbn-build-20051223.tar.bz2
. These are mostly a shameless rip off of Greg's system, but they're
customized to me. NOTE: this location is temporary. I'm not a
student anymore, and I think I'm losing the account in about two
weeks.
Next, I made two changes to the build order. The original book is
from Jeremy's home directory, version 20051216
http://linuxfromscratch.org/~jhuntwork/lfs-alphabetical/ .
Both changes are due to perl. First, I moved groff up. See this
thread http://linuxfromscratch.org/pipermail/lfs-dev/2005-December/054817.html
. I have to research this some more. It may be due to the DESTDIR
install I was doing. I don't know why Jeremy or Chris wouldn't have
seen this. Next, sed was moved up. Here's an earlier ICA report:
--- iter1/usr/lib/perl5/5.8.7/i686-linux/CORE/config.h 2005-12-22
01:42:14.000000000 -0800
+++ iter2/usr/lib/perl5/5.8.7/i686-linux/CORE/config.h 2005-12-22
10:44:54.000000000 -0800
[...]
@@ -24,7 +24,7 @@
/* LOC_SED:
* This symbol holds the complete pathname to the sed program.
*/
-#define LOC_SED "/tools/bin/sed" /**/
+#define LOC_SED "/bin/sed" /**/
[...]
--- iter1/usr/lib/perl5/5.8.7/i686-linux/Config_heavy.pl
2005-12-22 01:42:13.000000000 -0800
+++ iter2/usr/lib/perl5/5.8.7/i686-linux/Config_heavy.pl
2005-12-22 10:44:53.000000000 -0800
[...]
@@ -601,7 +601,7 @@
from=':'
full_ar='/usr/bin/ar'
full_csh='csh'
-full_sed='/tools/bin/sed'
+full_sed='/bin/sed'
gccansipedantic=''
gccosandvers=''
gccversion='4.0.2'
Seemed like a no brainer, but maybe sed is only used in the build.
Perl drives me crazy, so I didn't look into this further.
OK, on to the current results. I did ICA and Farce on 3 iterations.
Some directories were pruned. Here's the directories used (I don't
create the whole FHS tree):
$ ls ica/iter1
bin boot etc lib sbin usr var
I haven't done any research yet, but I'm attaching the ICA report for
1v2. With the exception of farce-extras (too big to move around), you
can see the results in http://students.washington.edu/dbnichol/lfs/ .
I'm going out of town in the morning, so I probably can't do any
research till the new year. I'd appreciate any comments people have.
--
Dan
REPORT.1V2
Description: Binary data
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
