...

Attendees 

• Danny Bernstein 

Agenda

Follow from last week: 

• Jared Whiklo 
• Peter Eichman 
• Joshua Westgard
• Doron Shalvi
• Longshou Situ
  
  
  

Agenda

...

1. Update on status of memento/sprint 2 work

   1. Remaining work
      1. creating mementos (partially complete)
         1. https://github.com/fcrepo4/fcrepo4/pull/1262 (

            Jira
            server DuraSpace JIRA serverId c815ca92-fd23-34c2-8fe3-956808caf8c5 key FCREPO-2617

            )
      2. getting listing of mementos
         1. FCREPO-2623
      3. getting mementos
         

         1. Jira
            server DuraSpace JIRA serverId c815ca92-fd23-34c2-8fe3-956808caf8c5 key FCREPO-2613

   2. Status of Mohamed Mohideen Abdul Rasheed's in progress PR: (https://github.com/fcrepo4/fcrepo4/pull/1262)
      
      1. Do we need a volunteer to bring this one over the finish line? 
   3. Status of https://github.com/fcrepo4/fcrepo4/pull/1250:
      1. Support versioning with versioned child ignored
      2. Is there anything that needs to be discussed as a group in order to move this one forward? (Longshou Situ)
   4. PR's to be merged
      1. https://github.com/fcrepo4/fcrepo4/pull/1261 (

         Jira
         server DuraSpace JIRA serverId c815ca92-fd23-34c2-8fe3-956808caf8c5 key FCREPO-2645

         )
      2. https://github.com/fcrepo4/fcrepo4/pull/12621260 (

         Jira
         server DuraSpace JIRA serverId c815ca92-fd23-34c2-8fe3-956808caf8c5 key FCREPO-2617

         )

         -2621

         )'
      3. https://github.com/fcrepo4/fcrepo4/pull/1241 (just needs to be merged - all feedback it complete)

2. Follow-up discussion from last week: 

   1. Status of Doran's write-up regarding Fedora's strategy for permalinks
   2. Abandoning the single subject restriction: any feedback/illuminating discussion on the list?

3. Stuck thread issue resolved! (post-op)

   1. https://groups.google.com/d/topic/fedora-tech/lPY4QGacnMg/discussion
4. ...

5. Tickets In-Review

   Expand
   Jira server DuraSpace JIRA columns key,summary,type,created,updated,due,assignee,reporter,priority,status,resolution maximumIssues 20 jqlQuery filter=14307 serverId c815ca92-fd23-34c2-8fe3-956808caf8c5

   
   
   

...

1. Please squash a bug!

   Expand
   Jira server DuraSpace JIRA columns key,summary,type,created,updated,due,assignee,reporter,priority,status,resolution maximumIssues 20 jqlQuery filter=13122 serverId c815ca92-fd23-34c2-8fe3-956808caf8c5

   
   

2. Tickets resolved this week:

   Expand
   Jira server DuraSpace JIRA columns key,summary,type,created,updated,due,assignee,reporter,priority,status,resolution maximumIssues 20 jqlQuery filter=13111 serverId c815ca92-fd23-34c2-8fe3-956808caf8c5

   
   

3. Tickets created this week:

   Expand
   Jira server DuraSpace JIRA columns key,summary,type,created,updated,due,assignee,reporter,priority,status,resolution maximumIssues 20 jqlQuery filter=13029 serverId c815ca92-fd23-34c2-8fe3-956808caf8c5

   
   

Minutes

DCFUG Users Questions:

1. Some users would like to relax the apparent requirement in modeshape to repository url is the subject of all URLs
   1. ie people would like to be able to specify their own subject URIs.
2. Aaron Birkland:  does the upcoming memento implementation which abandons referential integrity make lifting the single subject restriction feasible?
   
   1. Jared Whiklo:  I don't think so.  All we are doing for memento is replacing the jcr reference with a URI. 
   2. So only the object is being updated from a jcr weak reference to a URI.
   3. Should we consider abandoning the single subject restriction  -  now would be a good time to do it.
   4. Let's raise it on the list.
3. Does loosening the single subject restriction break RESTful crawling? 
   
   1. Some users of fedora 3 are maintaining the mapping on their own.
4. Are we talking about maintaining the mapping in Fedora?  Possibly.
   1. Question drives at doubts about Fedora's strategy fro sharing URI's broadly.
   2. Most user's are not providing direct intenet access to the internet so they are using a proxy layer.
      1. API-X can do this work.
   3. It would be useful to do a write up on the strategy for permalinks.
   4. IPFS - interplanetary file system  - might be useful to look into for inspiration.
   5.  Doran from NLM will do a writeup.

Esmé Cowles will drive conversation at Samvera Connect re Fedora 5.0.0.

Having an easy to access, automatically generated one-click on GitHub would enhance Samvera's ability to test fedora 5.

Samvera will wait for a fedora 5 release candidate before looking at migration.

We should consider putting out an "Alpha" or "Preview"  as soon as the API alignment work is done for evaluation purposes.

Updates on the status of the Sprint 2 work:

Done:

• create timemap
• enable versioning
• proper headers on versioned resources

Remains to be done:

• creating mementos
• getting listing of mementos
• getting mementos

Requirements for merging into master:

• create an rdf memento
• retrieve an rdfs memento
• list memento
• enable versioning on a resource

There is still significant work to be done.  Danny Bernstein, Jared Whiklo are trying to push the work forward.  It would be good to have another sprint to wrap it up, but may not be strictly necessary.

What's going on with clustering and high availability in modeshape? 

We should not rely on single file store or single database - it would be good to clarify the strategy here.

Actions

1. Memento work
   1. Mohammed's FCREPO-1262 was a Work In Progress start. It needs work to finish it. Danny Bernstein will try to push on this work and get it finished.
   2. Jared Whiklo not much progress on FCREPO-2623 
   3. Need some help merging some PRs that are in final stages
2. Permalinks
   1. Concerns about Fedora URIs to be used in an abstract manner to the outside work. No changes from last week. Doron Shalvi to write up the issue and e-mail out to the Fedora listserv.
3. Stuck thread issue
   1. It was the internal audit module that writes internal audit events back to the repository. Discovered by walking through starting from Modeshape kernel implementation to get a handle on what actual happens when you send a PUT/POST to the repo. What is the chain of steps when Modeshape persists. For every write to the repo, 2 audit records are written. One for the resource and one for the parent. So if we take the audit writes out of the equation, and it worked perfectly. So have removed the fcrepo-audit dependency since Monday. Was able to run 2 batches of 500 newspaper issues each. Each issue creates a couple hundred Fedora resources. With no stuck thread issue. Indexing was faster. Switched over to using the fcrepo-audit-triplestore camel route in development and testing it now. 
   2. Is there a bug there that needs to be fixed, like an endless loop? Not an endless loop, just the volume of writes that occur. Could possibly have been resolved by increasing the Modeshape buffer to an incredible size. It appears that multiple resources created inside one parent may have caused some of the lock conditions. No definite issue found in audit, but the N^2 writes from using audit seemed a good place to look.
   3. What would be helpful is to create a JIRA issue that pulls together the issue you've had and the resolution you found and points to what you think might be the issue. The we have a record and if someone else runs into it, they can find it. 
   4. There are unanswered issues to the direct cause, it appeared to be due to user-level transactions. But it also seems related to lower-level transaction handling. This is not a crazy use of the repository. They are 16 page newspapers, so several resources, proxy resources and binary resources but this should be usable.
      
      

Actions

• Danny Bernstein will try to push forward on FCREPO-1262.
• Doron Shalvi will write up a description of the Fedora URI concerns and post to the Fedora listserv to start discussion.
• Peter Eichman will create a ticket to document the stuck thread issue that UMD experienced and how they have resolved it thus far.
• Bring the proposal to disable versioning using a read-only LDPCv to the Fedora spec editors
• Document UMD's custom authNZ Tomcat configuration (Peter Eichman)