RS03 requires a minimum of 15 or 16 roots if --no-bdr-defect-management is specified

[jpovixwm]

When calculating the RS03 layout, dvdisaster tries to find the smallest medium size that will result in at least 8 roots. However, even if the --no-bdr-defect-management CLI flag is specified, the function RS03Layout *CalcRS03Layout(Image *image, int target) still performs its checks using defect-management-enabled medium sizes:

dvdisaster/src/rs03-common.c

Lines 536 to 563 in 45ff2c5

	else if(get_roots(dataSectors, BD_SL_SIZE) >= 8) {
	/* Single layered BD */
	if (Closure->noBdrDefectManagement)
	lay->mediumCapacity = BD_SL_SIZE_NODM;
	else
	lay->mediumCapacity = BD_SL_SIZE;
	}
	else if(get_roots(dataSectors, BD_DL_SIZE) >= 8) {
	/* Double layered BD */
	if (Closure->noBdrDefectManagement)
	lay->mediumCapacity = BD_DL_SIZE_NODM;
	else
	lay->mediumCapacity = BD_DL_SIZE;
	}
	else if(get_roots(dataSectors, BDXL_TL_SIZE) >= 8) {
	/* Triple layered BDXL */
	if (Closure->noBdrDefectManagement)
	lay->mediumCapacity = BDXL_TL_SIZE_NODM;
	else
	lay->mediumCapacity = BDXL_TL_SIZE;
	}
	else {
	/* Quadruple layered BDXL */
	if (Closure->noBdrDefectManagement)
	lay->mediumCapacity = BDXL_QL_SIZE_NODM;
	else
	lay->mediumCapacity = BDXL_QL_SIZE;
	}

This means that the NODM variant of a medium size will only be chosen if the equivalent DM-enabled medium size yields at least 8 roots.
I've tried obtaining the maximum image sizes that will result in a NODM variant being chosen, and came up with this table:

Medium size	Max ISO sectors	Roots	Redundancy
BD_SL_SIZE_NODM	11408740	15	6.25%
BD_DL_SIZE_NODM	22817482	15	6.25%
BDXL_TL_SIZE_NODM	45635950	15	6.25%
BDXL_QL_SIZE_NODM	58271740	16	6.69%

However, I'm not sure whether this is a bug or intended behavior, as I know that using a redundancy lower than 20% triggers a warning message anyway, so maybe these redundancies of ~6.25% are too small to be useful anyway?

[speed47]

Thanks for your report!

You're right, the above code is incorrect, and yields sub-optimal results when defect management is disabled. This was not intentional.
That said, a redundancy of 6.25% is not really a good idea, but it's still better than nothing, i.e. as you said, sub-20% get a warning printed, and if the user ignores it, we'll make the assumption that they know what they're doing. If for some reason you have to burn only one BD and you happen to have 6% space remaining, it's still better to fill it up with parity data than not doing anything with it!

[speed47]

fixed by #80, thanks for your report (and review!)