Time/Place
This meeting is a hybrid teleconference and IRC chat. Anyone is welcome to join...here's the info:
- Time: 11:00am Eastern Daylight Time US (UTC-4)
- Dial-in Number: (712) 775-7035
- Participant Code: 479307#
- International numbers: Conference Call Information
- Web Access: https://www.freeconferencecallhd.com/wp-content/themes/responsive/flashphone/flash-phone.php
- IRC:
- Join the #fcrepo chat room via Freenode Web IRC (enter a unique nick)
- Or point your IRC client to #fcrepo on irc.freenode.net
Attendees
- A. Soroka
- Yinlin Chen
- Esme Cowles
- Jared Whiklo
- David Wilcox
- Benjamin Armintor
- Doron Shalvi
- Kevin S. Clarke
- Unknown User (bbpennel)
- Diego Pino Navarro
- Bethany Seeger
- Andrew Woods
Agenda
- PUT with server-managed triples... fail or ignore?
- 4.5.1 Release planning
- One blocker:
- Release Testing Procedures
Bugs are beginning to pile up
- Fedora Specification updates
- Messaging SPI
- Atomic Batch Operations - name? BatchOps? Bag-o'-Ops? OpSack? AtomicOp?
- CRUD
- Resource Versioning
- Binary Fixity Checking
- Authorization
- Updates on release status of fcrepo-camel and toolbox?
- ...
Status of "in-flight" tickets
Ticket Summaries
Please squash a bug!
Tickets resolved this week:
Tickets created this week:
Minutes
- PUT with server-managed triples... fail or ignore?
- when you have a resource that you want to update with a PUT request (body being RDF, that doesn't not include server managed triples) the current logic tries to remove the triples that are not in the body of the RDF - throwing a 4XX error. Question - we not require the server managed triples be included in the body of RDF request. Use case is to update the resource w/o having to know the server managed triples, since we can't change them anyways.
- Esmé Cowles reminded us that there is a preferred header ```handling=lenient;received="minimal"``` should do the trick.
- Why is the default behavior taking server managed triples into consideration in the first place? Should it be the other way around? Lenient default - and have a strict option? Use case: one may want the full graph in a case where other's are editing at the same time - for conflicts.
- Esmé Cowles has used this recently - would be surprised if not implemented, since it's worked for him.
- Esmé Cowles will test this (the lenient and received=minimal)
- Confirms that PUT with received=minimal does do the requested behavior - updates the user triples.
- Esmé Cowles to update the documentation for this
- Should work for Stefano's use case.
- Andrew Woods - Is there any situation that should require putting the server managed triples back on the server? Would this capability help in some way? ETags should serve the same purpose.
- Concern about what it means if someone tries to write server manage triples that are wrong? Persist anyways?
- if you're trying to change the value on a server managed triple, you should get an exception. (including implicit removal for triple not specified in request).
- LDP spec says HTTP PATCH is for merge of data, and to use HTTP PUT for replacement.
- HTTP PUT for replacement - but with server managed triples it's more complicated.
- The delete and recreate use case is the one to figure out how to work with here.
- Want to be careful of situations with server managed triples that change w/o user knowing (Indirect Containers, for example).
- Two ways to do this stuff
- PUT with received=minimal
- SPARQL update via PATCH
- is it worth someone times to investigate a different behavior here, or are things good as is?
- handling=strict; received="minimal" may give same response, which would probably be a bug.
- Benjamin Armintor - to add comments to
- 4.5.1 Release planning - getting a release out would be beneficial
- one blocker -
- what do we think the expected behavior should be in this situation (a batch operation)?
- a link to an ACL that is prefixed with transaction context ?
- a link to an ACL that is not prefixed with transaction context?
- a link to an ACL that points to non-accessible uri?
- no ACL link header?
- behavior different depending on whether or not the ACL is created w/i the batch request?
- auth is a core feature and if not working, we can't really release. - what do we think the expected behavior should be in this situation (a batch operation)?
- Any other tickets in before we put out release? None put forward
- Who can work on release?
- Benjamin Armintor - Release Manager
- Esmé Cowles and Jared Whiklo happy to help out
- one blocker -
- Bugs are beginning to pile up