• We are at around 5000 users
• There is a signup performance problem related to the fact that the time it takes to add a node to another node is exponentially related to the number of sibling nodes of the node to be added
  • BCM is researching this: first step is to check out JackRabbit^?'s addNode() impl, see if there is anything obviously silly going on.
• Todd did some analysis and found that 6% of foxmarks files in Cosmo are invalid
  • is the problem Cosmo or Foxmarks
  • some probably due to the "partial put" problem which is now solved
  • some though are not just partial files, they have weird stuff like driver names in them
  • more research is required

Scooby Demo

• Do we wait for Cosmo?
  • At the scooby meeting the day before the consensus was yes, but that was based on the assumption that Cosmo 0.3 completion is far in the future
  • What if it weren't? BCM thinks he can reduce the estimated time if we put off Foxmarks fixes till like .3.1 (or maybe the Foxmarks issue is real easy)
  • upshot: bcm to research these possibilities, decision is on hold
• Terms of Service what should they be?
  • we need to have clear language that data integrity is not guarunteed
    • but we won't actually wipe the data every night/week - more interesting to see how stuff scales.

Scooby & Cosmo in the Wild

• People are downloading the bundle and using it, and checking out Scooby
• Also are using other clients with Cosmo like Sunbird and Evolution
  • Issue: these clients do not make it clear how to setup with a CalDAV server (eg they don't say that the URL you provide must be a PREEXISTING calendar collection) so we would like to fill this gap
  • BCM to provide documentation? Or at least start the page?

Scooby Release

• Devs are wrapping up with docs and comments
• QA is testing
• Mikeal is concerned that all these commits in Cosmo since scooby was tarred up will make QA confusing since Cosmo will hvae changed as well as Scooby between testing iterations
  • We can provide Mikeal with the same snapshot he used for the first iteration of testing
• Pieter is concerned about the trickiness of signing up for a Scooby account
  • the confusion is centered around the fact there is no such thing as a Scooby account - it's your Cosmo account
  • you have to sign up on cosmo, then go to Scooby to log in
  • Pieter wants this automatic, which is good, but not feasable for this release
• We feel scooby-demo must be running and available and ready before we announce Scooby 0.1
  • Running released scooby 0.1 will be a new instance for Jared to create
  • Jared proposes using Server-bundle, try to install it as an admin would.

Misc Cosmo

• The current Cosmo+Tomcat bundle will not be maintained
  • People who want the whole bundle will get Scooby as well whether they want it or not.
  • OTOH, People who want only Cosmo with more control, will have to set up their own tomcat, also their own jackrabbit, and use the (as yet poorly documented) WAR files to get Cosmo working on top of that.

More Scooby Release - Release Steps

• Pieter has a bunch of steps on landing page he can do before release...
• QA will vet the release candidate and approve a release
• Bear will cut a release and make available for download
• Jared will turn off the redirect on the default port at Scooby-demo and put up new instance -- this may be a week or more from today
• Some content needs to be written by various people, associated with landing page or Scooby docs or other
• We make a final cleanup on landing page after all these pieces are in place and test the experience of a person who hears about Scooby