Date: Fri, 3 Mar 95 09:26:59 -0500 From: crtb@helix.nih.gov (Chuck Bacon) To: questions@FreeBSD.org Subject: CDROM usable for building kernel? Message-ID: <9503031426.AA22803@helix.nih.gov>
next in thread | raw e-mail | index | archive | help
I just received O'Reilly's 4.4BSD manual set, with accompanying CDROM
of 4.4BSD-Lite. To find out if there's much code in common with
FreeBSD2.0R, I decided to compare /cdrom/4.4BSD-Lite/usr/src/sys with
/usr/src/sys. So I cobbled a perl script which asks the central
question: for each file in the /usr/src/sys tree, is there a matching
file in the /cdrom/4.4BSD-Lite/usr/src/sys tree which is identical?
If so, it's a candidate to become a symlink and save me some disk space.
Well, I was disappointed. Only 11 out of 602 files had identical mates
on the /cdrom. Following a sneaking suspicion, I found that if I
replace `cmp $a $b` in the script with `diff $a $b` and checked for an
output of two or less lines, I had 87 out of 602 files which I could
replace with symlinks.
A single line has been added to all of the source files, for RCS.
It looks like:
* stdarg.h,v 1.5 1994/08/02 07:39:09 davidg Exp
I'd like someone with a 2.0R FreeBSD CDROM, who is tracking -current
or 2.1dev., or any other variant, to run this script, and on a wider
tree than /usr/src/sys, perhaps /usr/src. I'd like to know about
how much disk space an updated distribution will take, and how that
grows over time. Obviously, I could build a 2.0R kernel easily with
no disk space, other than symlinks (and binaries of course :-).
I guess what I'm interested in is the dynamics of a static CDROM,
and the rate at which it becomes obsolete.
At the risk of derision for making such grungy code public, here's
the script I ran. Rewrite to suit.
=======================
#!/usr/local/bin/perl
# src-comp.pl
# Compare the CDROM and my /usr/src/sys directories, making a table of
# similarities and differences.
$cdrom = '/cdrom/4.4BSD-Lite/usr/src/sys';
$disk = '/usr/src/sys';
@diskdu = `cd $disk; du`; # Get names of all directories
for (@diskdu) { s/^\d+\s+\.\/(.*)\s+/$1/; } # Keep only the names
pop (@diskdu); # Discard "Total" line
for $d (@diskdu) { # For each directory, get all the plain files
$ddir = "$disk/$d";
$cdir = "$cdrom/$d";
chop (@ls = `ls $ddir`);
for $file (@ls) {
next if -d "$ddir/$file";
if (-r "$ddir/$file" && -r "$cdir/$file") {
@result = `diff $ddir/$file $cdir/$file`;
if (@result > 2) {
push (@nohits, "$d/$file");
} else {
push (@hits, "$d/$file");
print "HIT $d/$file\n"; # These 'prints' are volatile.
print @result; # I want to see that alleged RCS stuff
}
}
}
}
open (R,"> report") || die "Can't write report: $!\n";
print R "Hits\n";
for (@hits) { print R ' ', $_, "\n"; }
print R "\nChanged files\n";
for (@nohits) { print R ' ', $_, "\n"; }
close (R);
}
=======================
Chuck Bacon - crtb@helix.nih.gov
ABHOR SECRECY - DEFEND PRIVACY
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9503031426.AA22803>