dodecahedron wrote:Inertia, are you sure that there should be no red block when recording in DAO mode?
i seem to recall that i've burned CDs in DAO that had the last block red, and apart from that seemed completely error free!
My experience has been that of Inertia's: burn in TAO = red final block, burn in DAO = green final block. However others have sworn to experiences like that described by dodecahedron: a disc burned in DAO shows a red final block, but in all other respects shows absolutely no errors (including bit-wise comparisons of the files). So, I am willing to call the CDSpeed behavior a bug. It is either an actual functional bug (assuming the reports by others are accurate), or it is, if you will allow me to coin a phrase, a "UI design bug" - too many average users are mistakenly lead to believe that something is actually wrong with their CD-Rs because of confusion over TAO and DAO.
dodecahedron wrote:hey, cfitz, you mention here (and elsewhere) the md5summer program.
as far as i can tell, this program is used for creating MD5 checksums, not for verifying disc integrity.
can it be used for verifying disc integrity too? how?
Yes it can, in essentially the same way that the CRC option in CDCheck allows verification.
The idea behind both of these methods is to calculate a set of unique signatures for each file that is to be burned in a CD compilation. These signatures (32-bit CRC-32 for CDCheck, 128-bit MD-5 hash for MD5summer) are compact representations of the bits that make up each file, and are designed to change dramatically if even a single bit in the original file changes. Thus, if any file is corrupted by a failing disc, a subsequent calculation of the signature will result in a different signature that does not match the original. This allows you to detect the corruption.
To make practical use of this, you use the program to calculate a signature for every file that is to be burned on the CD-R, and record those signatures in a file. Both CDCheck and MD5summer automate this capability. You then include this signature file in the compilation, and burn the complete CD-R. Your CD-R now contains the actual data files and the additional signature file that lists each data file's signature at the time of burning. At any later time you can re-calculate the signatures of the data files stored on the CD-R, and compare those recalculated signatures to the original signatures. Any difference means that a file has been corrupted. Again, both CDCheck and MD5summer provide automated methods for recalculating and comparing signatures.
One problem with this method in general, of course, is that by the time corruption has been detected, it may be too late to recover. However, it does give one assurance that all is well if the signatures do all match. To detect problems before they become uncorrectable, you need to use a program like CDSpeed’s disc scan or quality check that can detect correctable C2 errors before they become uncorrectable errors. If a disc’s C2 errors significantly increase over time, it is an indication that the disc is degrading and that those C2 errors may eventually become so numerous and severe that they transition into uncorrectable errors and thus, file corruption.
Why would one want to use MD5summer instead of CDCheck? The MD-5 hash is a 128-bit signature, four times as long as the 32-bit CRC-32 signature used by CDCheck. As such, it is exponentially more robust at detecting changes than CRC-32, similar to the way that longer encryption keys lead to exponentially more difficult to crack encryption. There is a very
small but finite chance that a file could be corrupted in such a way that the CRC-32 code remains the same, and thus the corruption would go undetected. The chance of the same thing happening for an MD-5 hash is, for all practical
Is the additional strength of the MD-5 hash a good reason to choose MD5summer over CDCheck? Probably not, at least not for the typical CDRLabs reader. For all but the most stringent requirements, the strength of CRC-32 should be more than sufficient. Also, calculating MD-5 hashes is computationally intensive, so MD5summer takes more time to calculate the signatures and verify the files. Finally, CDCheck has extra CD-R specific comparison and recovery features that MD5summer does not include. However, I do like to mention MD5summer as an alternative and/or adjunct for those who are truly paranoid.