- On Wednesday March 3rd we will need to meet elsewhere, the tea room has been reserved by another group.
- Software development cycle: let's have a live deployment of the parts of the software we've been working on that is accessible for the meetings. We could have a cron job on whichever server is hosting it that checks out the projects nightly/weekly, builds them, and deploys them.
- There are authorization and authentication issues to be discussed.
- There are indexing issues to be resolved.
What is indexed, what is cached, what is left in outside data stores that must be connected to and queried out-of-band. We need to consider making sacrifices in indexing for response-time, there should probably be incremental indexing of some sort. If you assert an annotation, can it be indexed immediately? Where are the latencies for depositing annotations in the network? (dependent on network and not on content of annotation?) (the data in the annotation may have implications on length of analysis after triage)
- James has added to the use case scenarios: Use Case Scenarios
MeetingUser: Bob Morris | User: Zhimin Wang | User: Maureen Kelly | User:Donna Tremonte | User: Paul J. Morris | User: David Lowery | User: Jonathan Rees | User: Chinua Iloabachie | User:
Maureen proposes making a block at the beginning of each meeting to firm up an agenda for the meeting, and designates a person to handle the audio/video arrangements for the next week. Bob Volunteers for next meeting.
Software development life cycle
Maureen proposes a standing demonstration machine with a current functional demonstration of all FP components, one purpose being weekly demonstrations of activity. Public access OK. Bob notes that people's work may be out of sync, and API's for one component may not weekly be stable for other components to use from the cutting edge. Zhimin proposes using Maven for managing components. Probably needs multiple branches in SVN and availability of stable components.
Maureen proposes a nightly cron job to build all in trunk and deploy the project, with email to developers if build fails. This identifies build failures, documents the build process, and documents the deployment process. Needs deployment to several machines.
Looked at some synonomies and chresonomies, discussed. Some points came up:
Implications for character data annotation as assertions are made about the description of species in new synonyies. Annotations about concerns about data sets, rather than anotations about data sets. Also, annotations about missing data. Analogy to quality control annotations?
Need a message to assert that data are missing?
Distinction between assertions about missing scholarship, and important new data? No engineering distinction, but relevant to humans?
Issue, you know about the origin of some data, you know about the current state of the data, but you don't know the full provenance chain (analogy to fine arts world (assertions about author of work based on the work, and assertions about the work from the history of its provenance)): Does the academic work on data provenance have this as a special case? (Also there's a WC3 incubator group on provenance http://www.w3.org/2005/Incubator/prov/ ).
Authorization and Authentication
Defer to next week 2010Feb10.
For next week
People to make lists of components to go into Maven, to discuss details.
Bob to report on MMI's interests in distributed data annotation.