[mercury-users] Aditi (was Re: Mercury in academic teaching?)

[Simon Taylor]

On 21-Nov-2006, J??rg Roman Rudnick <joerg.rudnick at t-online.de> wrote:
> I am interestested in the code - especially in regard of how to build 
> Aditi upon PostgreSQL.

This would probably be a lot of work.
 
> You mean 'runtime query optimization' would require extending the 
> PostgreSQL query
> optimizer to readaptability during query execution, since Aditi 
> 'queries' tend to be somewhat
> more complicated than SQL queries?

Yes.  mmc converts recursive Mercury queries into an iterative fixpoint
computation, and each iteration may have very different relation access
patterns.
 
> Do you expect it to be possible to extend the query optimizer in 
> cooperation with the
> PostgreSQL team? As far as I know, it is a part they put a considerable 
> effort into, so
> they might be rather interested...

Query optimization is not the only problem.

SQL doesn't have any support for passing arbitary relations into a query
without recompiling the query each time.  This can be kludged around by
storing all temporary relations with the same schema in a single relation
with a relation id attribute.  The expensive versions of Oracle, DB2 and
SQL Server can handle this better with table partitioning.

Simon.
--------------------------------------------------------------------------
mercury-users mailing list
Post messages to:       mercury-users at csse.unimelb.edu.au
Administrative Queries: owner-mercury-users at csse.unimelb.edu.au
Subscriptions:          mercury-users-request at csse.unimelb.edu.au
--------------------------------------------------------------------------