2021-07-13 Technical WG Agenda and Notes

normalizing unicode hyphens

"Implementors should be prepared to normalize some common invalid characters that may be found in ARKs copy pasted from processed text. For example, when pasting an ARK that was broken during line wrapping, a user may inadvertently introduce newlines and spaces that should be removed, or a variety of dash-like (eg, en dash) Unicode characters that should be removed or converted to hyphens."

GJ: "should be prepared" maybe too vague; perhaps this should be "best practice" or "it is recommended..."
ACTION: GJ will try to reword this
TC: implementors normalize; is this more for resolvers, or is it a purely a client-side issue?
JK: in some ways it's a publisher problem; but since we have little influence over them it may be best to play defensively
RM: similar situation with malformed PDFs
JK: and same with decades of malformed HTML
GJ: in the spec "normalization" is only currently mentioned only in terms of lexical equivalence; is it ok to leave it that way?

more diagrams (2) with ARK part names

See http://n2t.net/e/pub/ark.html#name-ark-anatomy-2

GJ: having two diagrams that are so similar is confusing

TC: I'd agree; I hadn't seen the check digit language before
JK: check digits are still optional but the convention is explained
GJ: are we too constrained by ascii art? maybe it would be good to have an actual diagram
JK: the IETF specs are generally meant to be renderable with plain text
TC: we might want to do a BNF grammar at some point
RM: the diagrams are too similar; the first should be higher level than the second; it seems possible to eliminate some detail from the first?
KH: agreed; or perhaps have more than two diagrams

new terms:

• first resolver
• resource responder
• metadata responder

See http://n2t.net/e/pub/ark.html#name-resolver-chains-2

TC: who is the audience for the resolver chain discussion; would be good to have some use cases
JK: example, EZID will go from using N2T as a Metadata Responder to being its own Metadata Responder
JK: an important use case for N2T retaining the ability to be a Metadata Responder is for namespace splitting
TC: yes, we have seen namespace splitting

recommend/exemplify link header with metadata response?

https://n2t.net/e/pub/ark.html#name-overview-of-the-http-url-map

KH: I like the link header example
TC: agreed; maybe use the full URL instead of relative URL
TC: would expect the "self" reference
JK: missing: converse recommendation of link header for resource retrieval

adding n2t.net as infrastructure; current statement is weak

https://n2t.net/e/pub/ark.html#name-finding-a-name-mapping-autho

non-normative pointer to ARK extensions at arks.org/specs/

https://n2t.net/e/pub/ark.html#name-enhancements-and-related-spe

RM: the arks.org/specs/ endpoint is ok
KH: agreed
TC: agreed
TC: nice to be able to see what the latest, up-to-date recommendation is
ACTION: JK to establish specs stub page at endpoint