Hi Scott, hi Henry,
Scott Stanton schrieb:
Ideally we'd hash the file contents, but that's going to be too slow to
do on every startup. Checking the expected size is a good compromise.
It's a fast sanity check that should catch the most common problems.
exactly: hashing is not an option because of performance. I just applied
the attached patch instead of your proposed one: (Just adding the -s)
Henry: Does this solve your problem? Can you try the PAR::Packer from
Subversion (http://svn.openfoundry.org/par/PAR-Packer/trunk)?
Cheers,
Steffen
--- script/par.pl 2008-01-07 22:44:49.000000000 +0100
+++ par.pl 2008-01-07 22:44:44.000000000 +0100
@@ -696,8 +696,13 @@
}x;
my $extract_name = $1;
my $dest_name = File::Spec->catfile($ENV{PAR_TEMP},
$extract_name);
- $member->extractToFileNamed($dest_name);
- outs(qq(Extracting "$member_name" to "$dest_name"));
+ if (! -f $dest_name or -s _ != $member->uncompressedSize()) {
+ $member->extractToFileNamed($dest_name);
+ outs(qq(Extracting "$member_name" to "$dest_name"));
+ } else {
+ outs(qq(Skipping "$member_name" since it already exists
+at "$dest_name"));
+ }
}
}
# }}}
