[urn] General grumble about incorporating other identifiers in to URNs

For many uses, we have defined URN namespaces that incorporate
particular previously-existing identifier systems.  Over the past couple
of months there have been two proposals to create URN namespaces that
incorporate multiple other identifier systems.  I expect that there will
be more.

My observation is that people proposing such namespaces generally have
not thought through all of the necessary considerations and the
committee should be on guard about such proposals.  In particular, the
proposers generally assume that embedding identifiers from any one
system into URNs is straightforward and needs no further specification.

In my experience, this is rarely the case.

Consider as a typical example the "isbn" namespace.  The registration of
version 2 of the namespace required considerable discussion of several
factors:

- ISBNs conventionally contain hyphens in places that are canonical for
  each assigned ISBN.  Should the hyphens appear in the URN?
- Should the hyphens be considered significant for lexical comparison?
- When the check digit is "X", how is that represented in the URN?
- The first, 10-digit, version of ISBNs is embedded in the second,
  13-digit, version of ISBNs.  Should lexical equivalence consider the
  embedded form to be the same as the original 10-digit ISBN?

Note that these factors are all specific to the ISBN system, and there
is no way to specify in advance all of the factors that will be
relevant for all future identifier systems.

Thus, any URN namespace that is intended to incorporate in the future
additional identifier systems needs to specify a robust process for
checking that the incorporation addresses all of the relevant
requirements.

It's not clear to me that this is possible without requiring a new
version of the namespace registration and its review by the committee.

Dale

_______________________________________________
urn mailing list -- [email protected]
To unsubscribe send an email to [email protected]