Discussion of how Matt and Bobby are coordinating data persistence
and what remoting framework to use
ideally a high-level API that abstracts away XMLHTTP reqeusts
Discussion of data formats
Could use iCalendar -- not that bad to write data parsing in JavaScript
Could use xCalendar or our own variant of iCalendar semantics in XML syntax.
Data format translations should be delegated to Cosmo whenever possible
This is mostly an implementation detail for transport -- for most purposes, an Event will be an object on either side.
The object API should be roughly compatible with iCalendar semantics unless we need to diverge for our custom features.
Licensing discussion
JSON is LGPL... Bobby will send the license to Sheila for examination
Zimbra
Do they support WebDAV? No hint of this although theoretically to drop in as an Exchange replacement they'd have to.
Don't support CalDAV...
Could Chandler share heterogeneous collections of arbitrary documents on Zimbra?
Only on Linux (specific builds only)
Merits further investigation as a platform (repository) for Cosmo
Certainly lisa should contact to talk about CalDAV at some point
Version hell
For Cosmo, we have 0.2, a branch from there for 0.2 patches, and the trunk has 0.3 work.
On cosmo-demo we have port 443 running 0.2 and port 8089 running continuous builds of 0.3 trunk.
On QA-cosmo server we run 0.2 only for now.
Chandler builds have account information that points to port 443 of cosmo-demo server.
Some debate about whether we've got patches installed on 0.2 (port 443) of cosmo-demo.
Morgen and Grant sometimes test against yet other builds of Cosmo installed on their own machines, because BCM is fixing bugs for them on the 0.2 branch.
Builds of Cosmo will be validated by Aparna before we upgrade the version on 443.
One proposal is that Aparna should only need to upgrade QA-Cosmo and run against that when we're validating a build of Cosmo and considering upgrading the main cosmo-demo port to that validated build. The rest of the time Aparna should save work by testing against cosmo-demo.
Do people have a decent understanding of what state Cosmo is at? we've just discovered it's not ready for dogfood prime-time, is that broadly understood?
Lisa should send email characterizing the robustness of sharing (restating the starred items in this conversation).
M6 should not be released until Chandler's sharing URL points to something that passes Aparna's smoke test.
Is it reasonable to fix the Cosmo bugs in time for M6? Not unless we know the complete list of bugs -- right now they're being flushed out by features being added in Chandler right up to the M6 deadline. So this puts M6 at risk.
Should we add another port with another version to cosmo-demo server? one that points to the latest build (unvalidated) on the 0.2 cosmo branch? There's an open ticket so, yes.
we need to know more clearly when we need to wipe Cosmo server data.
Sheila will create a page with the status of builds on cosmo-demo and will update that weekly (triggered by this meeting)