[plt-scheme] Is SchemeQL still being developed?

For the record, the Snooze documentation hasn't been updated since  
version 1.5 (the trunk is currently version 3.0 or thereabouts).  
Snooze is a little more sophisticated these days, particularly where  
the query language is concerned.

The failure to document the changes is largely my fault. I've  
committed an example file to demonstrate the basics until I've had a  
chance to whip my documentation monkeys into action:

    http://svn.untyped.com/snooze/trunk/src/snooze/example.ss

We've got lots of ideas for improving Snooze and we'd *really* like  
to push things towards a better data model and query language. For  
example, I'd like to try to eliminate SQL altogether by using  
PostgreSQL's PL/Scheme on the database side. For the moment, this is  
just one more of those projects that we've pushed onto the back  
burner while we earn our livings.

Cheers,

-- Dave

>> Should this be interpreted as Snooze implements the a relational
>> database in the algebraic sense (Date and friends aren't too
>> fond of SQL).
>
> Not as such.  For a start, there isn't as much implemented as we would
> like (no relationships, for example, though the query language
> supports joins).  Furthermore, I don't think the relational model is
> the one programmers actually want, for the same reasons that we don't
> use tuples (or lists, or vectors) as our sole data type.  The
> relational model is uniform, in that you can join any set of tuples to
> any other set of tuples and still have a set of tuples, but tuples
> themselves are a pretty inconvenient data type beyond a certain size
> (and moreso in Scheme than, say, Haskell).  We're more interested in a
> model defined over records, which would have slightly different
> semantics for joins.
>
> N.