Bug in md5sum (5.3.0) when checking lists of files?

[rmerks]

Dear core-utils developers,

Today I have used md5sum to produce checksums of the files of a DVD I just
wrote, on a Linux system (SuSE10.0):

cd /media/CDROM

find . -type f -print0 | xargs -0 md5sum > ~/sums.md5

Then I checked the resulting list of checksums against the original files:

cd original_directory/
md5sum -c ~/sums.md5

Much to my surprise md5sum found a lot of differences and could not find some of
my files, despite the fact that the DVD was an identical copy of the original
directory.

It appeared that the differences and absent files were in a directory named:

DLA_JAAP_JMP0.0150_4

while I have another directory named:

DLA_JAAP_MP0.0150_4

which contains almost the same set of filenames, and a number of other
directories with near identical names.

After I checked the files in which md5sum found differences "manually" (as
below) there appeared to be no differences at all:

md5sum -c ~/dvd6.md5 | grep -v OK > ~/dvd6_check.txt

cd original_directory
for i in `cat ~/dvd6_check.txt | perl -e 'while (<>) address@hidden(/:/); print
"$line[0]\n";}'`; do md5sum $i; done > ~/dvd6_2.md5

cd CDROM
for i in `cat ~/dvd6_check.txt | perl -e 'while (<>) address@hidden(/:/); print
"$line[0]\n";}'`; do md5sum $i; done > ~/dvd6_3.md5

diff dvd6_2.md5 dvd6_3.md5

(result: no difference)

This would suggest to me that md5sum -c reads the files in the checklist from
the wrong directory when directorynames look similar, perhaps because it is
truncating the names or something similar?

I look forward to your reply and I hope this problem (if any) will be fixed in
future releases.

Thanks in advance.

Best wishes,
Roeland