On 10/29/13 5:16 PM, Paul Hoffman wrote:
On Oct 29, 2013, at 9:03 AM, Patrik Fältström <paf(_at_)frobbit(_dot_)se>
wrote:
I think it is important to not restart discussions already held regarding
different requirements on identifiers, requirements that in turn lead to
various alternatives on how they are allocated, managed and resolved. I do
not think one can have one identifier that fits all. Instead multiple kind
of identifiers are needed. Because of requirements on uniqueness (absolute,
low risk of collisions or not needed at all), persistence, human
readable/understandable, whether allocation and resolution should be
designed for read (lookups) or write (allocation), what the identifier is to
be used for (see id/loc discussions).
Having sat through many of those discussions with Patrik 15 years ago: +1
And having chaired NSRG, + 1/2. That is- it's always fair to look at
new developments, but let's at least be aware of what was covered by
others and build on their success (or avoid their failures).
Eliot