2009 brought a lot of changes to DSpace and Duraspace…
1.5/1.5.1 as the year began brought
- Switch to Maven and Overlays
- XMLUI (Manakin) as well as JSPUI
Midyear 1.5.2 brought
- Sword update to 1.3
- Coocoon to 2.2
- Various fixes
- New Auth Systems
- New Translations
- Graceful resolver for URN
Also during 2009, they worked on some DSpace 2.0
- Technology preview for OR2009
- Some incremental development
- Backported service manager to 1.6
- DSpace 2.0 needs community support to move forward to complete.
Whats next….
- More services/components
- More generalised data model
- Straightforward migration
- Apply current dev/designs/community techniques
Issues
The new modularity/component aspect, leads to refectoring existing code into right building blocks for re-use in future. Will always be 90% done.
Data migration means updating database which in worst case the content store – but not likely. Using a data model means changing the cdode and evolving local procedures and authorisation.
Two UIS are expensive to maintain 0 double work vs mismatched features and neither universally accepted. JSPUI- outdated and XML a bit buggy and hard technically.
Development community is relatively small but dedicated. They go above and beyond to create the next version. THey are working to meet local needs.
Solutions.
- Maven from 1.5.x
- Refactoring and repackage
- Overlays – multiple overlays can be unpredictable
- Service Managers and components
- Incremental migration to date
- New Model alongside all old data
- Augment data as required
- Move to new code
UI Solution – Choose one?
- Inevitably XMLUI manakin is the choice going forward?
- Resolves differences
- Provide some way to migrate customised JSPUIs
- Subcommunities?
- Address existing problems
- Cookbooks in place
- or keep both and have them compete?
- Platform : rapidly changing area – factoring solution into service framework
- Dev Team
- – get more committers
- – less tech and specialists?
- – share more components
- – rely on adjacent teams
- – more commitment?
2010
New Service framework, probably OSGi direction.
- Used by Sakai and Fedora
- Proven in Eclipse
- It bundles dynamic loading/unloading and evolving dependencies
Governance
- Apache Model
- PMC -1.0.+1
- Majority for procedural issues
- Minimum of 3 +1 for release
- -1 can veto a code change
- Who are PMC? existing committers.
->
Dspace and Fedora to maybe merge to one solution in dspace2.0/fedora as turnkey solution?
Some Timelines…
1.6 oct/nov/dec window
OR2010 brings pressure to push to release next version, so 2nd/3rd quarters…
Some Questions
What about roadmaps for sword/lni ? History on sword is that as it improves, it rolls into the next version. Sword 3.0 has just been funded.
