[Debburn-devel] Burning corrupted data CDs with wodim

[Thomas Schmitt]

Hi,

> I get the rough idea of
> the backend (cdrecord), [...] (cdrkit), which
> then changed its name (wodim), but I don't know
> what libburn is to this.

libburn is an independent effort to operate
optical media. It does not share code with 
cdrecord or its clones.
The GUI program xfburn uses libburn
unconditionally. Brasero uses it optionally
instead of a cdrecord compatible program.
On top of libburn i maintain a cdrecord emulator
named cdrskin, and a combined ISO 9660 + burn
program named xorriso.

(I am not involved in cdrkit/wodim development.
 But this seems to be a general problem, so i
 am curious about its cause.)


> I should add that I successfully burnt an audio CD and that the DVD burning
> and reading seem to work like a charm.

If it is really systematic then it might be
about write mode TAO, which Brasero used.
There is a choice in xfburn "TAO" versus "SAO".
If SAO is in effect then we would have at least
a small technical difference.


> Anyway, I guess that's up to me to figure out.

Yes, i fear.

You should also avoid the GUI programs and work
with a ISO 9660 file on disk and the burn
backends directly.
The ISO file would be produced like:

  genisoimage -o /tmp/test.iso \
              -r -J -input-charset utf8 \
              /some/directory/with/a/few/hundred/mb/of/files

or you may copy a readble CD to disk:

  dd if=/dev/sr0 of=/tmp/test.iso


Then you can compare original cdrecord, wodim,
and cdrskin which all three understand the same
options. Like:

  prog=wodim
  $prog -sao -v dev=/dev/sr0 speed=4 \
       driveropts=burnfree fs=29m -data -nopad \
       /tmp/test.iso

(I added option -sao which was not in effect in
  http://bugzilla-attachments.gnome.org/attachment.cgi?id=162641
 and omitted tsize= which is unneeded if the
 input is a disk file rather than stdin.)


> Before proceeding to extensive testing, I need to know
> 
> - if CD-RW burning tests are relevant (I hope so...)

Except blanking and the physical differences of
the dye, CD-R and unformatted CD-RW behave the
same.
If CD-RW work reliably when CD-R do not, then the
physical differences would be a suspect and
indicate that the drive deteriorates or that the
CD-R media are poor.

> - if audio CD tests are useless,

If there are no audible faults, then the CD is
more or less ok. Audio does not have any
checksums. So it can use all 2352 bytes of a
CD sector as payload, whereas data sectors
normally contain only 2048 payload bytes.

> - if DVD burning tests are relevant (I guess the answer is no)

Not of interest if there is no indication of
failure.

Currently it is about gathering statistical
evidence that something is wrong with wodim in
particular.
I would rather bet that it is about -sao versus
-tao and/or about media-drive compatibility.

We'll see.


Have a nice day :)

Thomas