Attendees: Mischa, Andrea, Nicolas, Hannah, Mine, Romain

Notes:

• Andrea has done a first attempt at seeing a real world example in our schema https://docs.google.com/document/d/1gLDiLu410XPqB1egpiJJiRh3sPTG-KsbpN1p9OQb7rI/edit?usp=sharing 
  • Concern about the security logic of scopes vs path for e.g. copying a file
  • Since scope definition doesn't follow unix logic, might lead to complications
  • We have been following SciTokens path authorisation logic (see https://scitokens.org/technical_docs/Claims)
  • For WLCG typically have a flag that determines whether overwrite is allowed - commonly used
  • Our schema doesn't seem to match well with webdav 
  • We don't have a "metadata" authorisations concept, do we need one?
• Point made that difficult to follow fragmented discussions between lists
• 2kb limit
  • We asked on the OIDC R&E WG list about the origin of the apparent issue with tokens above 2kb
  • From Roland Hedberg "It’s a real world problem. And I think where it really hits is when the JWT 
    is part of a URL. Like when you have an id_token_hint in an authorisation request."
  • Including tokens in HTTP Headers seems to be OK
  • public archive http://lists.openid.net/pipermail/openid-specs-rande/Week-of-Mon-20190520/thread.html
• Not much energy from call participants recently
• Self imposed deadline to try to have final draft of schema document for September
  • TO-DO list per person perhaps
  • Hannah to liaise with Ian Collier to see whether we can have the pre-GDB 

Actions:

• Ask Brian and/or Jim Basney via email r.e. path semantics for storage permissions
• Ask Paul how DCache implements SciTokens in terms of scopes
• Hannah include summary of R&E mailing list r.e. 2kb limit (and link to public archive http://lists.openid.net/pipermail/openid-specs-rande/)
• Andrea to start thread to get feedback 
• Andrea to ask Fabrizio about tokens in URLs
• Hannah set up next call (first and second week of July)