Date
Attendees
Discussion items
| Time | Item | Who | Notes |
|---|---|---|---|
announcements: | Some discussion of need for "global" review of spec before release for review so that changes all cohere, more polished view, and can give good high-level Curtis and John spoke with some of repondants to poll. Comment from Pablo ? from Argentina that it is hard to find docs on ARKS (where to go - spec out of date, do I go to AITO?) He went to arks.cdlib.org – Need to clean this up quickly (part of outreach groups remit) (Pablo created Spanish-language ARK wikipedia page, is willing to translate FAQ as well) Roxana reporting on enhancements to their resolvers, etc. ; waiting on final spec for other changes | ||
(from last time) research handling of query parameters that have no explicit value (eg. ...??) by HTTP servers and reverse proxies. | Tom | No clear one place to go for clarity on behaviors of reverse proxy servers; Specs vs. was is coded; suggestion is to write a script and test against actual implementations; Suggestion that Tom will continue to monitor, he will try against top 3 proxies to see results | |
| question of collapsing ? and ?? | Greg: refers to Bertrands suggestion to use query term (non-empty) - why would we not? John: question as to whether appearance of English words in standardized query an issue (internationalization) (eg info) 2nd issue: historically IETF reluctant to do anything to standardize query string - although using ? or ?? in effect standardizing anyway Tom: 'meta'; ark-info, ark-meta; maybe also consider as optional method of support is via request-header rather than in query string; would provide greater latitude in describing methods of formatting response Mark: value also in having URL-like string that functions Greg agrees use cases for both techniques - string easier for "naive" implementation- question would this be an extension header (X-)? Tom yes - but style now for non-standardized without X quite common; also agrees, even if we do headers, useful to have query string John: given we will approve inflection supported, considering adding one with word - John would like to continue; could have language like ?info preferred, reserve? and ?? since that has been in spec for so long Curtis - prefers info, about - ? ok if words, options to add more operations in future John - we are moving to, at very least ? means same as ??, and ?? preferred Should be one query that gives basic information plus persistence info Sheila prefers word, okay with both Mark: has gotten feedback of questions about ? – looks strange to many for usual URLs; also implementations low; using word, making it "look" like query parameter would encourage greater use and implementation; re "both" - Might be confusing to have 2 ways of doing it; good with collapsing the formerly 2 ops (metadata + Persistence) into one combined op Roxana - most eager for clarification, decision so can proceed with implementation; for her ? is way to ask for info, but in web world, in URL, expect query behind it Tom finds ? problematic (Jetty strips it out, concerned other servers might do same) - seems more natural to have word; really happy if can add accept headers, they make routing decisions based on that; keyword in query plus accept headers would be very useful John - accept header good way to start; hybrid (both query string and header) Greg - definitely likes use of words; ? or ?? not obvious to outsiders what difference is - looks like typo - HTTP uses words in lots of places (method names, headers); query in URL makes it possible to bookmark it - strongly in favor of query term Tom: is the the time to dive into what metadata should be returned and what formats John suggests caution; perhaps converge on a few well-understood elements (who what when) then permit (any) other name-value pairs to prevent holding up this spec; make it a different working group's work Tom - lots of clients that make REST calls; try to get more detailed metadata where clients are asking for it Roxana - agrees with John's caution - eg what is the language for metadata (far more problematic than work in query string) ; Luxembourg uses 3 languages; John: need to support metadata better, avoid holding up spec for this, but still must be done as perhaps main goal for this group; Spec - generic description of services Mark - limited use cases that all ARKS must comply with; then have suggestions, recipes, reference to other document for other use cases Sheila should we write up minimum use case description John yes tech needs tightening it up to express bare minimum- language encoding richer metadata can happen in subsequent WG (this or other) activity -all | ||
| question of NAAN assignment corresponding to ISSN | John: Proposal to construct submission for request to NAAN to allow specification of use of ISSN; restrict to beginning with reserved initial letter; Sheila our experience ISSN not totally stable and persistent - worry about complications for NAAN maintenance Roxanne - -not an organization, an object managed by an organization; NAAN organization that manages object, NOT the object being managed BNF have ARKs for catalog records; records merge - re-direct old ARK to new one | ||
| when to open spec review to arks-forum |
Action items
- John Kunzewrite up summary of new take on inflections; capture consensus for all to review
- John Kunzewill also work on minimizing, clarifying metadata returned by inflections
- all email any further thoughts on ISSN as NAAN
