Atomicness of stopping to process Catalog Zones in section 5.1

[wtoorop]

This came from implementation feedback from our implementer @Koenvh1 oenvh1 In several places in the text we have something like: "* indicates a broken catalog zone which MUST NOT be processed (see Section 5.1)" But you detect such brokenness only while processing the catalog zones. @Koenvh1 was wondering if processing was atomic, so that all processing done up to the point where the brokenness was detected should be rolled back. I told him that that was not the idea and that the catalog zone would be considered broken from the point it is detected that it is broken. But maybe this should have been made explicit.

[peterthomassen]

@Koenvh1 was wondering if processing was atomic, so that all processing done up to the point where the brokenness was detected should be rolled back. I told him that that was not the idea and that the catalog zone would be considered broken from the point it is detected that it is broken.

That seems to be dependent on the (non-predictable) order of records e.g. in an AXFR, which seems suboptimal behavior (in terms of interoperability).

Alternatively, the zone could be evaluated for brokenness in one go (while reading it), and only be "processed" if valid.