P2: CVS/Subversion -- needs archival backup, not mirrored
P3: Wiki -- currently have mirrored but not archival
P3: archive of commit list
P2: WebDAV server for test/dogfooding
P2: Fix XSLT-generated content model documentation
P2: Get cheap (desktop?) linux box & KVM switch for Jed
P2: Free-text search of Wiki through Nutch -- needs update
Bugzilla configuration
P3: Add colour-coded diff viewing feature
P3: fix duplicate bug reporting
P3: Free-text search of source code through web interface (install Swish-e)
P3: Web hosting of auto-generated (by build?) epydocs
P3: IMAP server(s) for test/dogfooding
P3: Maintain blog hosting (support repository blogue, possibly additional) - would need to choose one engine that has appropriate licensing, features, and spam limitation...
P3: SSL connection to subversion
Possible: Get Mac for Stuart to avoid mac breaks? Lisa needs to investigate other fixes first
Possible: Tinderbox should be able to include tests which now fail due to "headless" state? We could make it work without a display, or could have benefits of having a real display. John's going to investigate whether this is a simple tinderbox environment variable problem.
Questions for design (someday) around sharing with both P2P and server-central models combined
Can somebody's item collections be shared in both models -- some in server-centric, some P2P?
Do we indicate to a user how each item collection is shared, to set their expectations?
Do we allow users to switch an item collection from one model to another?
Do we default to one of the two models?
How do we transfer the hub in the hub-and-spoke model, to a more available machine? Or do we configure the sharing circle to the right hub to start with?