In hacking up a Google Spreadsheets prototype, I've arrived a solution for connector
metadata that is the same as the Text File Connector's connector driven metadata
solution. I think an Amazon Simple DB solution lends itself to the same solution.
For anybody not familiar, a user creates a text file describing the metadata of the text
file, the connector consumes the metadata file and creates metadata for Teiid, which
exposes a dynamic VDB from the connector's metadata.
Eventually someone will want to model these sources in Designer to federate and virtualize
them. For text files the user has to currently hand model. We could create importers for
the different text file representations for Text, Spreadsheets, Simple DB, etc. Or,
perhaps the connectors could expose the metadata that they provide to Teiid through a new
API that a 'generic' connector metadata importer could consume?
~jd