Attendees

• Andrew Woods 
• frank asseg
• Benjamin Armintor
• Stefano Cossu
• Michael Durbin

General

• Indicates who took minutes - 
• Call-in: Google-hangout at: 
  • https://plus.google.com/hangouts/_/fedora-commons.org/f4-default?hceid=eW91cm1lZGlhc2hlbGYuY29tXzVlYzdpNXQ2Z282dTdidHI4aTVrbGJxOTUwQGdyb3VwLmNhbGVuZGFyLmdvb2dsZS5jb20.703tpqk1u0ampkfh15tiidvlb8

Agenda

1. Define issue
   1. Need for flexible content-modeling
   2. Need for repository resource validation
2. What proposals are on the table?
3. Short-term resolution? Long-term resolution?

Minutes

Content Modeling

Stefano

1. Use cases
   1. Assigning content model
      • Defining children and properties
      • This is provided (to some degree) by cnd
   2. Use a type to define behavior
      • Access policies
2. Desire to not allow mixins to be removed
   • This is offered by primaryTypes
3. There are limitations to cnd, as noted in the email thread
   • CND meets 90% of needs
   • Issue - wildcard properties
   • Issue - ranges of childern or properties are not supported

Ben

1. Wary of use case that making a primaryTypes and mixin immutable
2. F3 model, everything is effectively a mixin
3. Introducing concept of primaryType has significant implications
4. Could be constructive to put together use cases in functional terms
   • Use cases should avoid having implied implementation

Stefano

1. Other proposal is to start with more strict default primaryType

Frank

1. Need to ensure the ability to port F3 content models to F4

Andrew

1. What would the implications be if a primaryType could be user-configurable?

Mike

1. It would need significant experimentation

Stefano

1. Noted wiki use case: https://wiki.duraspace.org/pages/viewpage.action?pageId=34666309

Mike

1. Adding another mixin does not prevent users from creating more properties
2. We could immediately create a more restrictive cnd, changing default mixin

Frank

1. We do not want to make it difficult for new users to add content

Ben

1. Concern about impacts, for example, on WebDAV or sequencers

Frank

1. F4 currently creates nested folders for performance
   • Is there a risk if we change the default nodeType?

Test proposals

• Change CND to be more restrictive
• Sync/async validation of objects
  • Concern of the fact that invalid objects will already be ingested
• F4 proprietary content modeling rules (independent of CND)
• Change default primaryType from nt:folder
  • fedora:resource, leave it to the repo manager
• Allow user to provide their own primaryType

Actions

• Stefano Cossu to review/create use cases
• Stefano Cossu to create example CND
• Andrew Woods to explore implementation options to meet use cases