Cluster Limitation questions...

[Andrew Poodle]

I'm developing a proof of concept cluster, and would like some help and/or 
clarification of some of the points I've come across in the limitations.

No Text Indexes:  Does this mean that columns with datatype "text" cannot be 
indexed?

Can someone clarify what is meant by prefix indexes?  Is this where only one of 
several indexed columns in a table is intended to be used?

The 8 k row limit, I recognise this is probably to keep down traffic between 
nodes..  am I right in thinking if a field is changed to a TEXT or BLOB 
datatype, then those records aren't affected by the 8k limit? Is this 8k limit 
the actual data size of the row contents?

(I'm starting to have scary thoughts about some tables that currently have 
indexed TEXT cols.. to convert to a different format would bump them over the 8k 
limit per row, but to leave as is, would affect indexing)

oh fun :)

the foreign Key construct is ignored:  can I stll call cascades thu a query on a 
  table with a PK, and see the affects cascade thru the FK children, either 
delete or set to null whatever..

I'd like to do this at db level and no application layer.

thanks in advance...

a