[Debburn-devel] Repeatable images

[G. D. Chase]

On Mon, Feb 09, 2009, Steve McIntyre <steve at einval.com> wrote:
> On Mon, Feb 09, 2009 at 02:12:05PM -0800, G. D. Chase wrote:
> > It would be nice (at least I think so) for the disc images to be
> > repeatable.  Currently they are not.  See below for what I mean.
> >
> > $ genisoimage --version
> > genisoimage 1.1.2 (Linux)
> > $ mkdir mydir
> > $ echo filecontents > mydir/myfile
> > $ genisoimage mydir > firstimage
> > $ genisoimage mydir > secondimage
> > $ diff --brief --report-identical-files firstimage secondimage
> > Files firstimage and secondimage differ
> >
> > G. D. Chase
>
> What exactly changes from one the next, then? I'd only expect things
> like timestamps in the ISO header, and we have patches submitted to
> allow you to specify those on the command line. If that's all, then we
> can easily give you consistent images. :-)

$ diff --text firstimage secondimage | strings
makes it appear that there are time stamp differences, so I'm guessing
your expectation is right on.

$ mkdir mydir
$ echo filecontents > mydir/myfile
$ tar --create mydir > first.tar
$ tar --create mydir > second.tar
$ diff --brief --report-identical-files first.tar second.tar
Files first.tar and second.tar are identical

If tar is consistant, why isn't genisoimage?  I don't know a whole lot
about the ISO 9660 format . . .  Why is there a timestamp?  Is it
required (by the format)?  What ramifications would there be for leaving it out?  If one were to leave it out, would the space where it would
have gone have to be padded with nulls, or is the format a tagged format
instead of a fixed-position one?  Is this the right list to be asking
these questions?

I found the patches you referenced
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498151
which mentiones four timestamps: creation_date, modification_date,
expiration_date, and effective_date.  Could you point me to some
documentation which describes these timestamps?

Also, I just discovered isoinfo.  Does anyone think that it would be
good to include these timestamps in the 'isoinfo -d' output?

G. D. Chase