Date
8 September 2016
Attendees
Documentation Sent
https://docs.google.com/drawings/d/1Kv065ev5j3xcEfeIpsW_6hzXW8d9Z7xdH2hx4UYXH44/pub?w=960&h=720
Github: Github project page
PHYDO Data Model
PHYDO asynchronous storage interactions
Working mockup from Open Repositories 2016 Demo
Features Roadmap August-December 2016
Questions
Please add questions for Adam here
Discussion items
Time | Item | Who | Notes |
---|
| | | |
| | | Sufia/CC merger - Carrick has been on the calls with the arch. group, so he's probably been filling you in on that |
| | | PCDM - 2.0 may not happen |
| | | Fedora performance - we're looking at switching the direction of memberships to make Fedora faster |
Notes
PCDM 2.0 may not happen
recent discussion between islandora and hydra camps on IRC, islandora people wanted to fork 2.0 bc hydra people aren't receptive to their needs
- Centering around filesets -- islandora people want to be able to attach a file to an object and leave it at that, Hydra people want to use filesets - idea behind 2.0 is that that would be required
- Call last week: decided filesets are OK, as long as they're not required
Stefano Cossu discussion - want to move those around and have them as other objects?
Having the technical metadata as part of this fileset with object not a problem; becomes a problem when you try to index that
- Not a problem, but would have to basically push up all of the indexed portion of the metadata into the fileset layer; write your own fileset indexer class; parsing of xml to get those values out
- Need to add ... something, xml file in the header -- Esme Cowles -- this files describes this other file-- describedBy?
- Within the file need additional predicates to id the metadata file
Not a blocking issue, just have to add some work into this
Our hanging MODS off off the object directly is an islandora thing -- to do what Hydra wants to do, we'd put that into a fileset
Describes predicate would be pointing to the object itself -- Esme Cowles is the person to contact for this (he wrote the PCDM 2.0 spec -- there are other people who've had this same use case)
- If you want that MODS xml indexed you can index it as the fileset or index it as the object - doesn't think this matters
- How does discoverability of this object vs. fileset affect this decision?
Adam suggests treating it as a binary -- putting in a fileset (fits runs) - you've be able to do things like generate a checksum
The fileset idea is the representation of one original binary (so the original idea would contain the derivatives as well)-- Julie says that these technically are three different binaries we're getting from MDPI
In filesets - custom predicates (you can identify the different pres/prod/access) to define the different filesets
Julie: prob don't want to make fileheader changes on received files; Adam: not modifying original file, it's an ldp thing -- http head response --
Sufia/CC merger
document prior to HydraConnect to outline all of the decisions
much of Sufia can go back into Curation Concerns; other things: flippable features that can be turned on and off
This is being driven by Hydra in a Box work from Stanford