CD Speed doc (ScanDisc) wrote:...Audio CD. Unlike data CD's, the yellow areas are uncorrectable...
I think this says it all. This can only mean that uncorrectable C2 errors are measured.
If only it was so straightforward...
This just brings us back to what exactly is meant by "measured" and what is meant by "C2", which I think (at least as far as my reading of the interchange goes) was the crux of the initial disagreement between Inertia and Hitokiri and later spath. And even here, despite the enmity, I don't think there was disagreement on at least some of the basic principles. Everyone seems to agree that areas of the disc containing sufficient errors as to be uncorrectable at the C2 level are detectable and reportable to some degree. I think Inertia and spath even agree that within those areas, the exact number of bytes in error (or stated another way, the error type as in E32 vs. E42*, etc) can not be determined. Which still doesn't tell us what exactly CD Speed (or CD Doctor, for that matter**) is calling a C2 error and what the units are: is it E12 errors? E22 errors? E32 errors? E12 + E22 errors? E12 + E22 + E32 errors? Measured in absolute numbers of bytes with errors? Measured in bytes with errors per second? Measured in absolute numbers of blocks with errors? And so forth and so on...
One thing to toss in here is that EDC/ECC codes are generally able to detect more errors than they can correct. However, that doesn't mean that the EDC can properly classify how many errors occurred. Nor does it mean that the EDC can detect all errors. For example, a simple even/odd parity bit EDC scheme will detect one-bit errors, but can't correct any of those one-bit errors. It can also detect three-bit, five-bit, seven-bit, etc. errors, but can not distinguish any of these errors from any other, including one-bit errors. Moreover, it can't even detect two-bit, four-bit, six-bit etc. errors. The EDC/ECC used in CDs is more sophisticated than a simple parity bit, but the salient point remains the same: characterizing the behavior of these is not as straightforward as we would like.
So is this all useless? Nope. I will reiterate what I said earlier in this thread:
cfitz wrote:However, this doesn't invalidate the quest to characterize media and media/drive combinations. Certainly relative assessments of media quality are fair game. We may find that media X does consistently better than media Y when tested under identical conditions on the same drive. We can then say that media X is superior for those conditions on that drive. And we may find that media X does consistently better than media Y under a number of different sets of identical testing conditions. Then, we can expand our conclusion to state that media X is generally superior to media Y (keeping mind that there may always be exceptions for specific test conditions).
By the way, Miyuri's notes the readme.txt file of CD Doctor state the same thing - a consumer drive and CD Doctor (or WSES or CD Speed - take your choice) does not make a dedicated, specialized test system, so the results obtained thereby shouldn't be seen as absolute measurements but rather relative comparison data.
Now for a short detour to address a specific comment in cdrfreak’s post:
cdrfreak wrote:Reporting correctable C2 errors does not have much practical use.
It depends on one's application. If one is writing CD player software, no it does not have much use, unlike the reporting of errors that are uncorrectable at the C2 level as you rightfully point out. However, if one is trying to gauge the quality of one's CD-R burns, then yes it does have practical use. Forgive me as I again quote myself from earlier in this thread:
cfitz wrote:A disc may have no uncorrectable errors but have so many correctable errors that it has no margin for further degradation. A little more aging or a small scratch or two may send the error rates beyond the ability of the error correcting codes to correct. Is that the sort of disc to which you want to entrust your valuable data for long term storage? It may work right this moment, but how about tomorrow?
And back to the more general topic:
So, do we yet have a definitive explanation of CD Speed or CD Doctor's measurement and reporting methodology? No. Heck, for all I know there might not even be a single definitive answer since it may depend on what each drive reports from its internal EDC/ECC processes and that may vary from chipset to chipset. So, I will finish with a quote from Ian and one last quote from myself, since I think both are still valid:
Ian wrote:The only thing I'm taking from this whole thread is that CD Speed's documentation does not say what its showing in its tests and no one really knows for sure.
Anyone who knows the intimate details of CD Speed's testing methodology, please share with us!
* I don't think 'E42', 'E52' etc. is standard nomenclature but rather a logical extension of the original definitions. Accordingly, from here on out I will use 'E32' only, and use it in its original definition of an error consisting of 3 or more
bytes with errors passed to the C2 level of error correction.
** CD Doctor does plot is graphs with the label "CPS" which I take to mean "characters per second" - thus, I am assuming the figures are in bytes with errors per second. I suppose it might also mean "corrections per second", which would lead one to conclude that it is only reporting the correctable errors. But this remains speculative at this point.