This is a reminder that now that we have gone live with SNAC, the persistent identifiers are now persistent, which is to say, we have an obligation to make them so.
In this regard, we need some policies and procedures with respect to splits, merges, , and . Perhaps other issues as well, but those crop into mind.
My view is that it is best to brand all records that split into other records as cancelled with
ark blah blah
for each "split out" CPF. Any link to an ARK that is "cancelled" will be by definition, ambiguous.
How about from the new CPFs resulting from the split:
ark blah blah
??
CPF does not have a "cancelled-split" but that would probably be good, as otherwise the burden of knowing what happened falls on the .
Further, in also does not have a value for "cancelled-split" though one can use human-readable language in to describe, e.g, cancelled because of split (or some such).
Further, I would do the same for merges, though perhaps it is not necessary to exercise this level of caution. But herewith
cancelled with
cancelled plus , e.g. cancelled because of merge.
ark blah blahark blah blah
[Leaving out the details that follow the same pattern as a split].
Alternatively for "merge", one is chosen to "live" and merged is cancelled, with appropriate
This is a reminder that now that we have gone live with SNAC, the persistent identifiers are now persistent, which is to say, we have an obligation to make them so.
In this regard, we need some policies and procedures with respect to splits, merges,, and . Perhaps other issues as well, but those crop into mind.
My view is that it is best to brand all records that split into other records ascancelled with
for each "split out" CPF. Any link to an ARK that is "cancelled" will be by definition, ambiguous.
How about from the new CPFs resulting from the split:
??
CPF does not have a "cancelled-split" but that would probably be good, as otherwise the burden of knowing what happened falls on the.
Further, in also does not have a value for "cancelled-split" though one can use human-readable language in to describe, e.g, cancelled because of split (or some such).
Further, I would do the same for merges, though perhaps it is not necessary to exercise this level of caution. But herewith