Use cases
---------------
Exchange an entire DB via an exchange file
Return a result set from a data query in the exchange format
Synchronize two existing DBs using exchange syntax 

Desiderata
----------------

Participants; Spencer, Thau, Hutchison, Maffei, Jones

a. Thau: Layered complexity: Baseline of things you can represent, and then layered complexity
b. Nick: shouldn't be intimidating to domain scientists
    Andrew: but is it for domain scientists, really?
c. Readability: important for data longevity, somewhat less so for domain scientists
d. Domain scientists should be comfortable with the format
e. Viv: Should be automatically extractable from spreadsheets or whatever
f. One-off spreadsheet data is part of data for exchange format (but might be difficult)
g. Types of data to be handled:
   1) tabular/relational
   2) raster of data: e.g., vegetation index on plot
   3) vector data: e.g., bird migration path, geometries of plots or regions (point, line, space)
       -- sometimes the vector is the emasured value, others it provides context for the measured value
   4) binary blob: read off instrument, interpreted by specialized software
   5) image data/audio data (camera sensors)
h. Andrew: Ability to include a reference to a binary file in an external location (to represent binary information that is part of a row for example)
i. Nick: Effectively transportable: How transportable should this be?  Are there size limits for transportable chunks?
j. Thau: Does verbosity matter?   Thau: No. Computers can deal with it. 
k. Thau: Separation of semantics from data stream in order to avoid duplication (annotation-based approach)
l. Clear implementation path: Needs to be clear about how to create serializations (not too many options for how to represent the same information)
m. Needs to be supportable by tools that can mediate interaction with scientists (e.g., conversin tools)
    -- needs to be supportable by extant tools (no specialized dev needed) (that scientists are already using)


Requirement
-------------------
Should handle tabular data, both normalized and denormalized to various degrees
Should be able to package binary streams as part of serialization format


Additional comments from afternoon session
-------------------------------------------------------------
-- should have demonstrable implementations along the way
-- Wiezcorek -- extensibility of Darwin Core
-- localization should be considered as well

Next steps
---------------
-- JWGODMS mailing list
-- next meeting: organized and open, SONet funded?
-- summarize the Desiderata and Requirements work in a document