Page tree
Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 8 Next »

Time/Place

Time: 10:00am Eastern Time (US)

URL: https://lyrasis.zoom.us/my/fedora

Meeting ID: 812 835 3771

Dial by your location:
888 475 4499 US Toll-free
877 853 5257 US Toll-free
855 703 8985 Canada Toll-free
0 800 031 5717 United Kingdom Toll-free

Find your local number: https://zoom.us/u/ad6Xb7q3ia

Attendees

  1. Tammy Allgood Wolf
  2. Melissa Anez
  3. Chris Awre 
  4. Thomas Bernhart
  5. Danny Bernstein 
  6. Robert Cartolano 
  7. Aaron Choate
  8. Sayeed Choudhury
  9. Stefano Cossu
  10. Jon Dunn 
  11. Karen Estlund
  12. Dan Field
  13. Raman Ganguly
  14. Jennifer Gilbert
  15. Babak Hamidzadeh
  16. Neil Jefferies
  17. Mark Jordan
  18. Danny Lamb
  19. Rosalyn Metz 
  20. Este Pope 
  21. Scott Prater
  22. Robin Ruggaber 
  23. Tim Shearer
  24. Erin Tripp 
  25. Andrew Woods 
  26. Dustin Slater (star)
  27. Jennifer Vinopal 
  28. Evviva Weinraub
  29. Jared Whiklo
  30. David Wilcox
  31. Wei Xuan
  32. Maurice York 
  33. Laurie Arp
  34. Robert Miller

Agenda

Topic

Lead

Financial Update

  • Budget overview and quarterly report
Laurie

Fedora 6 design decision

David

Previous Action Items

  • Governance Group - Review subgroups and organization before end of June. Jennifer Vinopal Rosalyn Metz 
  • Maurice York and Robin Lindley Ruggaber to ask their teams how things are currently structured on disk to compare against the OCFL spec.
    • Robin - the structure on disc is the way the Samvera stack structures them (for Fedora 4). Did not receive a satisfactory answer for Fedora 3. Not sure if we will move that forward. But we expect to be able to migrate. 

Notes

Budget Overview and Quarterly Report:

(document was attached to email sent to group by David Wilcox)

Previous years not available due to differences in fiscal year accounting between Duraspace and Lyrasis. The complexity is too great. Robert's key number is the Net Assets. That keys him in on what resources he has to accomplish objectives.

Lyrasis views a positive revenue as a strength. Building up a reserve provides stability and flexibility and is a symbol of health. 

Q: Since membership dues aren't a steady income, how are they handled?

A: Invoices go out at the beginning of the year. Institutions generally pay promptly. Lyrasis recognizes the income at 1/12th of the total per month. 

Q: How do you managed delayed payments? Or Institutions that don't renew?

A: We initially expect the revenue to come in, as it is understood that it sometimes get delayed. However we quickly follow up with organizations. If people indicate they are not going to renew there is CEO/leadership engagement. We call the institution to understand the impetus for not renewing.  We use this information to try to identify trends across the  the 10 community supported organizations under the Lyrasis banner. This allows us to get ahead of the trend for other related products.

Q: When we look at this sheet how should we consider the total assets?

A: It follows accounting practices, so it includes reserves. 



Fedora 6 Design Decision

Core issues and implementation options

David introduced the topic, which was a summary of the document linked immediately above.

Q: (MY) Why not just create a default "Version 0" for everything on the backend?

A: (DW) If we did this .... we still need to figure out what to do with the content later. Future changes would need to be versioned or else the "Version 0" directory would be overwritten with changed content. Versions in OCFL are supposed to be immutable.

A: (RM) OCFL assumes you are being intentional. Fedora does not assume the user is being intentional. It allows for a broader use case of individuals not thinking about what happens when they update content. 

Q: (SP) Does the Fedora tech team have a preference from their technical point of view?

A: (DW)They do, but we're hesitant to introduce that this early in the discussion. We will share that later in the discussion, but we want to prejudice the discussion.



(SP) Repository owners will have 2 main concerns:

1) Can I rebuild my repository from disk?

2) Is my content versioned or not?

If technical details can be overcome, there is no reason not to version.

(RM) My concern about auto versioning is that it will create a lot of storage. This will become a point that the community will have concerns about.

(TS) Cloud costs for storage could be prohibitive with auto versioning.

(AW) Every time you create a new version, you get a new manifest. So there is a text file that gets larger with each version. On the extreme end, we have seen spreadsheets with 35,000 versions. 

(EP) What are the objects that most often get large number of versions?

(AW) The above scenario was that there was a parent container that got version every time a child was modified or added. The act of creating a version in OCFL is the act of preservation. Indicating that it is ready for preservation. In Fedora the act of creating a version has been more about wanting to not lose a previous version.

(Mark (in chat)) Should "auto" not be determined further up the stack than Fedora? In other words, Samvera and Islandora, and other business logic layers, determine whether all changes result in new versions, or some contextual/configuration logic that makes this curatorial decision on behalf of the user. It is conceivable that multiple applications may be using Fedora, and some may want auto and some may not.
 

(DW) This is something I worry about too. If Fedora is too rigid in how we deal with it, it could dissuade other applications from adopting Fedora. We want to make it as easy as possible to adopt Fedora 6. So we want to limit how much work others will have to do to adopt Fedora. So, can we come up with a solution that aligns as much as possible with OCFL, but not make the change too complicated to adopt?

(DW ran through the options in the above document)

(TS) In the mutable head scenario, is it a given that auto versioning is an option? Not a default?

(AW) I think these are orthogonal topics. If you have auto versioning on at the application or Fedora level, mutable head wouldn't exist.

(DB) Having the system just auto version makes it more simple to not have a mutable head, but if we have a mutable head it makes versioning easier.

(SP) <asked about having unversioned content in the OCFL structure>

(AW/RS) You must not include other objects in the OCFL object root is the specification. See the last sentence in the draft https://ocfl.io/draft/spec/#object-structure

(RS)  I think the language in the specification should be change from "must not" to "should not". I think it is ok to have content in a directory that hasn't been versioned.

(DW) Whatever we choose, it is going to be a compromise. What the decision gets down to is: where should unversioned content live? We want to hear from this group on what the community thinks, and what the best way forward for Fedora?

(TS) What is the urgency/timeline for this decision?

(AW/DB) There is a sprint that starts November 4th. We would like to have the decision a week previous to that date.

(MY) The view that Fedora enforces a particular viewpoint has been an issue in the past, so we should consider that when we decide what Fedora does. The flexibility is important.

(DW) Time is up today. David will follow up with DB and AW after the call and put together some material to help make the decision. The core issue seems to be where to store unversioned content.


Action Items 


  • No labels