Apr. 16, 2004

Agenda

metadata registry - are native XML database technologies the way to go? I'm currently evaluating eXist and will soon look at Apache Xindice, bdbXML and anything that y'all think looks appropriate (hopefully open source, or free to UCSB).
What lineage queries does the lineage engine need to handle?
what API is acceptable to the user for capturing metadata (building XML)?
Is Tomcat/Java servlet architecture acceptable.? eXist can 'hide' behind Tomcat so the user doesn't have to mess with it.
What is the set of web services that ES3 will provide?
Seems like we should develope the lineage engine first, then prototype clients with Python GUI, then production clients in a more full featured GUI toolkit like Java/Swing (don't cringe Frew) or the like.

What are our plans with Thetus? I don't think we actually decided what lies ahead with this.

[Debbie] Sometime ago we talked about re-organizing the internal web. I created a "mock-up" based on input at that meeting at http://essw.bren.ucsb.edu:8081/webdev/. On Friday, lets take a quick look and see if we want to switch over or leave things, as is.

different classes of md
- required md
- recommended md
- standard scientific md
- non-standard scientific md

we need to distinguish between attributes and relationships

we need to distinguish between programs and processes

in Rajspeak, this is analygous to workflow and workflow invocation
a process creates files, not programs
the metadata object for a program describes the program's parameters and settings

a process doesn't need to know all the files it has created
a file should know which process has created it
the md for a file should contain this 'backward pointer' to the object that created it
- the lineage can be created from these backward pointers
- the ES3 server may aggregate these lineage clues so it can efficiently perform lineage queries