[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Repeatability of mkisofs?



Michael Shell wrote:

I've got an interesting one. I keep track of the md5sums of all the images
I burn. Now, it just so happened that I had deleted an image file that
I later needed. So, I rebuilt a new image via mkisofs from the exact same
directory structure used to create the old image. Now, the two images had
the exact same byte count, but, much to my surprise, they differed with
respect to their md5sums! Why is this?

Old versions put command line info in the image, creation date is preobably there, and quite likely the atime has changed on parts. I use "sumdir" to embed md5sums in the directory(s).

I used almost the same command line each time:

mkisofs -dvd-video -o image.img directory

The only difference was in the name/path of the output file. I would
expect mkisofs to be deterministic in that the same input data will
produce an identical output result (for the same version of mkisofs,
in my case this is 2.01). Several possibilities for the differing output
files come to mind:

1. mkisofs somehow makes use of a random number generator
2. mkisofs is embedding or using the current time in the output
3. mkisofs is embedding the output filename/path or the command line as
  it was invoked within the created image

It would also be good to know if there is a combination of command
line options which will result in identical images with repeated calls
to mkisofs (and if not, maybe this would be a good feature to add to
mkisofs in the future).


 Thanks in advance for any info that sheds light on this,

 Mike Shell





--
E. Robert Bogusta
 It seemed like a good idea at the time



Reply to: